CN1508511A - A Calibration Method for Structured Light Vision Sensor - Google Patents

Abstract

The invention belongs to the technical field of measurement and relates to the improvement of a sensor calibration method in structured light three-dimensional vision detection. The steps of this method are setting the target—fixing the sensor—collecting images—extracting target coordinates—projecting structured light—extracting the coordinates of points on the structured light strip—extracting the coordinates of any point on the structured light strip, using double cross-ratio invariance Find the three-dimensional object coordinates of the point—find R and t—find a, b, c, d. The calibration point of the method has high precision, and the number can be arbitrarily many, and the measurement precision of the three sensors to be calibrated is high, and is suitable for on-site calibration.

Description

A Calibration Method for Structured Light Vision Sensor

technical field

The invention belongs to the technical field of measurement and relates to the improvement of a sensor calibration method in structured light three-dimensional vision detection.

Background technique

Among many visual inspection methods, structured light three-dimensional visual inspection is widely used in the measurement of workpiece integrity and surface flatness; automatic inspection of microelectronic devices (IC chips, PC boards, BGA), etc.; soft, fragile Detection of parts; detection of three-dimensional shapes of various molds; visual guidance of robots, etc. Structured light visual inspection technology has been more and more widely used in industrial environments due to its characteristics of large range, large field of view, fast measurement speed, easy extraction of light strip images and high precision. The effective calibration method of structured light vision sensor model parameters has always been an important research content. At present, there are mainly the following methods: one is the sawtooth target method. Duan Fajie et al. in the article "A New Method for Structural Parameter Calibration of Line Structured Light Sensor" (Journal of Instrumentation, Vol.21No.1, 2000) and Liu Fengmei et al. "(Measurement Technology, No7.pp.3~6, 1999) stated this method. The principle of this method is shown in Figure 1. This method is mainly to project the light plane onto the jagged target, so as to form some bright spots on the tooth edges as calibration points. This method has the following problems: firstly, the extraction precision of the image points is low because the tooth edges are easy to reflect light. Second, because of the light intensity distribution of the bright spot itself, it is difficult to strictly correspond to the bright spot in the space with the electronic theodolite and the bright spot in the image. Third, due to the limited tooth edges, the number of calibration points obtained is small. Another method is the cross-ratio invariant method. In 1995, Xu Guangyou of Tsinghua University proposed to use the principle of cross-ratio invariance to obtain The method of calibrating points, but the implementation is too troublesome. In 1999, D.Q.HUYNH wrote the article "Calibration of a Structured Light Stripe System: A Novel Approach" (Calibration a Structured Light Stripe System: A Novel Approach), International Journal of Computer Vision, Volume 33, Issue 1, 73-86 Page, 1999 (International Journal of Computer Vision, Vol.33 No.1, pp.73-86, 1999) also proposed a calibration point acquisition method using the principle of primary cross-ratio invariance. Coplanar points, each group of 3 points and collinear, use these 12 points to obtain the three-dimensional coordinates of 4 points on the light plane by using the cross-ratio invariance for the calibration of the line structured light vision sensor. But the common problem of both is that the number of calibration points is still small, so the calibration accuracy needs to be improved

Contents of the invention

The technical problem to be solved by the present invention is to provide a high-precision structured light vision sensor calibration method suitable for on-site calibration, further improve its calibration accuracy, and improve the convenience of its engineering application.

The technical solution of the present invention is: a calibration method for a structured light vision sensor, characterized in that,

(1) Set the target, the target is composed of two planes, the angle θ is 45°≤θ≤135°, there are pre-set feature points on the two target surfaces, and the target surface is one of the following structures:

A. Target A. There are two columns and three rows of raised black rectangular blocks on the target surface. The number of rectangular blocks in each row is 2 to 100, and the distance between the rectangular blocks is 10 to 200mm. Four square blocks of each rectangular block vertices are feature points;

B. Target B has circular holes arranged in a matrix on the target surface, the number of rows and columns of the circular holes is 3 to 100, and the center of the circular hole is used as the feature point;

C. Target C, there are crosshairs arranged in a matrix on the target surface, the number of rows and columns of the crosshairs is 3 to 100, and the intersection points of the crosshairs are used as feature points;

(2) Fix the sensor and the target at a certain distance, and ensure that the light plane emitted by the laser in the sensor intersects the two target surfaces;

(3) Turn on the power supply of the CCD camera in the sensor, collect an image of the calibration target, and store it in the computer;

(4) For target A, extract the two-dimensional image coordinates (X _i , Y _i ) of all vertices of the black square on the target and the corresponding three-dimensional object coordinates (x _wi , y _wi , z _wi ), and store them in the computer;

(5) Turn on the power of the laser, so that the projected structured light plane intersects with the two planes of the target to form two structured light strips;

(6) Extract the two-dimensional image coordinates (X _Li , Y _Li ) of the intersection point between the structured light strip 5 and the edge of the black square on the target body, and use the first-order cross ratio r(A _i , B _i , C _i , D _i ) =r′(A _i ′, B _i ′, C _i ′, D _i ′) calculate and store the three-dimensional object coordinates (x _wLi , y _wLi , z _wLi ) corresponding to the image coordinates (X _Li , Y _Li ), and store them, the formula r(A _i , B _i , C _i , D _i ) means the cross ratio of A _i , _{B i ,} C _{i , D i} ’) represents the cross ratio of A _i ′, B _i ′, C _i ′, D _i ′;

(7) Take any point on the structured light strip, extract its image coordinates (X _Li , Y _Li ), combine the three-dimensional object coordinates (x _wLi , y _wLi , z _wLi ) obtained by the first cross ratio, and use the double cross ratio Invariant formula r(D ₁ , D ₂ , D ₃ , D ₄ )=r'(D ₁ ′, D ₂ ′, D ₃ ′, D ₄ ′) to obtain and store the three-dimensional object coordinates of the point;

(8) Repeat the work of (7) to obtain the three-dimensional object coordinates and corresponding two-dimensional image coordinates of any number of points on the structured light plane;

(9) Using the vertices obtained in (4), the pair

$\left[\begin{array}{c}\mathrm{\ω}\·x\\ \mathrm{\ω}\·Y\\ \mathrm{\ω}\end{array}\right]=\left[\begin{array}{cccc}f& 0& 0& 0\\ 0& f& 0& 0\\ 0& 0& 1& 0\end{array}\right]\left[\begin{array}{cc}R& t\\ 0^T& 1\end{array}\right]\&Center\; Dot;\left[\begin{array}{c}{x}_w\\ {\mathrm{the\; y}}_w\\ z_w\\ 1\end{array}\right]$ Perform calibration to obtain R and t, where ω represents a proportional constant, f represents the effective focal length of the CCD camera, R is a 3×3 rotation matrix, and t is a 3×1 translation vector;

(10) Use the points on the structured light plane obtained in (6), (7), and (8) to calibrate the following formula a· _xw +b· _yw +c· _zw +d=0, and obtain a, b, c, d, where (a, b, c, d) represent the coefficients and constant terms of the light plane equation.

The advantage of the method of the present invention is: the method is based on the principle of double cross-ratio invariance, and the structured light visual sensor is calibrated. Compared with the existing calibration method of the structured light visual sensor, the calibration points of the method have high precision and the number can be arbitrary The measurement accuracy of the three calibrated sensors can reach 0.1453mm, 0.1524mm and 0.1496mm respectively. This method is suitable for on-site calibration.

Description of drawings

Figure 1 is a schematic diagram of the principle of the existing sawtooth target method. In the figure, 1 is a sawtooth target, 2 is a laser projector, and 3 is a CCD camera.

Fig. 2 is a schematic diagram of the principle of constant cross ratio.

Fig. 3 is a schematic diagram of the structure of a calibration target in the method of the present invention. In the figure, 4 is the calibration target body, and 5 is the structured light strip.

Fig. 4 is a schematic diagram of the structure of the second calibration target in the method of the present invention. 6 in the figure is the calibration target body, and 7 is the structured light strip.

Fig. 5 is a schematic diagram of the structure of the third calibration target in the method of the present invention. In the figure, 8 is the calibration target body, and 9 is the structured light strip.

Fig. 6 is a physical diagram of the structured light vision sensor, in which 10 is the sensor housing, 11 is the CCD window, and 12 is the laser exit.

Detailed ways

The method of the present invention will be described in further detail below. The method of the invention calibrates the structured light vision sensor for the first time based on the double cross-ratio invariant principle. The method can obtain any number of high-precision three-dimensional object point coordinates required on the structured light plane, and is used for sensor calibration. First, the principle of double cross ratio invariance is briefly explained. Under the perspective projection transformation, the length and the ratio between the lengths can be changed, but the ratio between two ratios about the length is invariant. As shown in Figure 2, there are three non-coincident straight lines A _i B _i C _i (i=1, 2, 3) on the plane π ₁ , and the straight line D ₁ D ₂ D ₃ intersects these three straight lines at points D ₁ and D respectively ₂ , _D3 . Through the perspective projection center o, their images on the plane π ₂ are A _i ′B _i ′C _i ′ and D ₁ ′D ₂ ′D ₃ ′ respectively. According to the theorem of perspective projection, a straight line is still a straight line after perspective projection transformation. Therefore, points A _i ', B _i ', C _i ', D _i ' (i=1, 2, 3) are collinear. The cross ratio of collinear four points is defined as: $r(A,B,C,\mathrm{D.})=\frac{\mathrm{AC}}{\mathrm{CB}}/\frac{\mathrm{AD}}{\mathrm{DB}}---\left(1\right)$

According to the principle of invariant cross ratio under perspective projection transformation, the following formula holds:

r(A _i , B _i , C _i , D _i )=r′(A _i ′, B _i ′, C _i ′, D _i ′) (2)

Under the condition that the coordinates of points A _i , B _i , C _i and the coordinates of points A _i ′, B _i ′, C _i ′, D _i ′ are known, the coordinates of D _i can be obtained by using formula (2). In this way, the coordinates of points D ₁ , D ₂ , and D ₃ can be obtained respectively.

Take any point D ₄ on the straight line D ₁ D ₂ D ₃ , and the perspective projection point corresponding to point O is D ₄ ′. Using the principle of cross-ratio invariance again, we have:

r(D ₁ , D ₂ , D ₃ , D ₄ )=r′(D ₁ ′, D ₂ ′, D ₃ ′, D ₄ ′) (3)

The coordinates of points D ₁ , D ₂ , D ₃ , the coordinates of the corresponding image points D ₁ ′, D ₂ ′, D ₃ ′ and the coordinates of point D ₄ ′ can be obtained from formula (2) with the same cross ratio. Under known conditions, the coordinates of D ₄ can be obtained by using the cross-ratio invariant formula (3) again. By analogy, the coordinates of any point on the straight line D ₁ D ₂ D ₃ can be obtained. In this paper, the above-mentioned characteristics of the cross-ratio are called double cross-ratio invariance.

Mathematical model of structured light vision sensor.

The mathematical perspective projection model of the structured light vision sensor consists of two parts: the perspective projection imaging model of the camera and the light plane equation of the structured light.

According to the principle of camera pinhole imaging and the principle of perspective projection transformation, the perspective projection imaging model of the camera can be described as follows:

$\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; x</span>}\\ \mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; Y</span>}\\ \mathrm{<span\; class="notranslate">\; \&omega;</span>}\end{array}\right]<span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; f</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; f</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\end{array}\right]\left[\begin{array}{cc}\mathrm{<span\; class="notranslate">\; R</span>}& \mathrm{<span\; class="notranslate">\; t</span>}\\ {\mathrm{<span\; class="notranslate">\; 0</span>}}^{\mathrm{<span\; class="notranslate">\; T</span>}}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; \&CenterDot;</span>\left[\begin{array}{c}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ {\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ {\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\\ \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

in $R=\left[\begin{array}{ccc}{r}_1& r_2& r_3\\ r_4& r_5& r_6\\ r_7& r_8& r_9\end{array}\right]$ is the rotation matrix, $t=\left[\begin{array}{c}{t}_x\\ t_{\mathrm{the\; y}}\\ t_z\end{array}\right]$ is the translation vector, and ω is the scale factor, which is a constant.

The equation of the light plane equation can be described as: a x _w + b y _w + c z _w + d = 0 (5)

Formula (4) and formula (5) constitute a complete mathematical model of structured light vision sensor. According to this model, the calibration of the structured light vision sensor is divided into two steps: one is the calibration of the camera imaging model parameters R and t, and the other is the calibration of the structured light plane equation. Solving R and t in formula (4) requires at least 6 non-coplanar 3D world points and their corresponding 2D image points; determining formula (5) requires at least 3 non-collinear 3D object points on the light plane . In order to improve the calibration accuracy, it is often necessary to obtain more high-precision calibration points.

The concrete steps of the inventive method are as follows:

(1) Set the target, the target is composed of two planes, the angle θ is 45°≤θ≤135°, there are pre-set feature points on the two target surfaces, and the target surface is one of the following structures:

A. Target A. There are two columns and three rows of raised black rectangular blocks on the target surface. The number of rectangular blocks in each row is 2 to 100, and the distance between the rectangular blocks is 10 to 200mm. Four square blocks of each rectangular block vertices are feature points;

B. Target B has circular holes arranged in a matrix on the target surface, the number of rows and columns of the circular holes is 3 to 100, and the center of the circular hole is used as the feature point;

C. Target C. There are crosshairs arranged in a matrix on the target surface. The number of rows and columns of the crosshairs is 3 to 100, and the intersection points of the crosshairs are used as feature points;

(2) Fix the sensor and the target at a certain distance, and ensure that the light plane emitted by the laser in the sensor intersects the two target surfaces;

(3) Turn on the power supply of the CCD camera in the sensor, collect an image of the calibration target, and store it in the computer;

(4) For target A, extract the two-dimensional image coordinates (X _i , Y _i ) of all vertices of the black square on the target and the corresponding three-dimensional object coordinates (x _wi , y _wi , z _wi ), and store them in the computer;

(5) Turn on the power of the laser, so that the projected structured light plane intersects with the two planes of the target to form two structured light strips;

(6) Extract the two-dimensional image coordinates (X _Li , Y _Li ) of the intersection point between the structured light strip 5 and the edge of the black square on the target body, and use the first-order cross ratio r(A _i , B _i , C _i , D _i ) =r′(A _i ′, B _i ′, C _i ′, D _i ′) calculate and store the three-dimensional object coordinates (x _wLi , y _wLi , z _wLi ) corresponding to the image coordinates (X _Li , Y _Li ), and store them, the formula r(A _i , B _i , C _i , D _i ) means the cross ratio of A _i , _{B i ,} C _{i , D i} ’) represents the cross ratio of A _i ′, B _i ′, C _i ′, D _i ′;

(7) Take any point on the structured light strip, extract its image coordinates (X _Li , Y _Li ), combine the three-dimensional object coordinates (x _wLi , y _wLi , z _wLi ) obtained by the first cross ratio, and use the double cross ratio Invariant formula r(D ₁ , D ₂ , D ₃ , D ₄ )=r'(D ₁ ′, D ₂ ′, D ₃ ′, D ₄ ′) to obtain and store the three-dimensional object coordinates of the point;

(8) Repeat the work of (7) to obtain the three-dimensional object coordinates and corresponding two-dimensional image coordinates of any number of points on the structured light plane;

(9) Using the vertices obtained in (4), the pair

$\left[\begin{array}{c}\mathrm{\&omega;}\&CenterDot;x\\ \mathrm{\&omega;}\&CenterDot;Y\\ \mathrm{\&omega;}\end{array}\right]=\left[\begin{array}{cccc}f& 0& 0& 0\\ 0& f& 0& 0\\ 0& 0& 1& 0\end{array}\right]\left[\begin{array}{cc}R& t\\ 0^T& 1\end{array}\right]\&Center\; Dot;\left[\begin{array}{c}{x}_w\\ {\mathrm{the\; y}}_w\\ z_w\\ 1\end{array}\right]$ Perform calibration to obtain R and t, where ω represents a proportional constant, f represents the effective focal length of the CCD camera, R is a 3×3 rotation matrix, and t is a 3×1 translation vector;

(10) Use the points on the structured light plane obtained in (6), (7), and (8) to calibrate the following formula a· _xw +b· _yw +c· _zw +d=0, and obtain a, b, c, d, where (a, b, c, d) represent the coefficients and constant terms of the light plane equation.

Example

The actual design of the structured light vision sensor is shown in Figure 6. 10 is the housing of the sensor, 11 is the window for the CCD to observe the scene, and 12 is the exit port of the laser. The positions of CCD and laser in the sensor are different, and their parameters are also different.

According to the steps described above, three specific sensors were calibrated using the target (θ=90°) shown in Figure 3 .

Sensor one:

·Camera perspective projection imaging model parameters

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; 0.135832</span>}& \mathrm{<span\; class="notranslate">\; 0.171788</span>}& \mathrm{<span\; class="notranslate">\; 0.975725</span>}\\ \mathrm{<span\; class="notranslate">\; 0.843045</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.537361</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.022753</span>}\\ \mathrm{<span\; class="notranslate">\; 0.520408</span>}& \mathrm{<span\; class="notranslate">\; 0.825670</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.217816</span>}\end{array}\right]\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{c}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 61.97129</span>}\\ \mathrm{<span\; class="notranslate">\; 8.464447</span>}\\ \mathrm{<span\; class="notranslate">\; 601.586335</span>}\end{array}\right]$

·Structured light plane equation coefficient

[a b c d]=[-0.000769 0.056868 -0.584231 10.404217].

The repeatability test errors of the calibrated sensor reconstructing the three-dimensional object coordinates are:

$\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.021</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.0345</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.0033</span>}\mathrm{<span\; class="notranslate">\; mm</span>}$

The RMS error of the distance between two points in the measurement space is: E _RMS =0.1453mm.

Sensor two:

·Camera perspective projection imaging model parameters:

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; 0.212029</span>}& \mathrm{<span\; class="notranslate">\; 0.151030</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; 965522</span>}\\ \mathrm{<span\; class="notranslate">\; 0.713005</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.699573</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.047147</span>}\\ \mathrm{<span\; class="notranslate">\; 0.668332</span>}& \mathrm{<span\; class="notranslate">\; 0.698418</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.256015</span>}\end{array}\right]$

$\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{c}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 68.096325</span>}\\ \mathrm{<span\; class="notranslate">\; 6.559659</span>}\\ \mathrm{<span\; class="notranslate">\; 609.599409</span>}\end{array}\right]$

·Structured light plane equation coefficient

[a b c d]=[-0.00848 10.022128 -0.271455 7.8725557].

The repeatability test errors of the calibrated sensor reconstructing the three-dimensional object coordinates are:

$\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.027</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.0445</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.0036</span>}\mathrm{<span\; class="notranslate">\; mm</span>}$

The RMS error of the distance between two points in the measurement space is: E _RMS ＝0.1524mm

Sensor three:

·Camera perspective projection imaging model parameters

$\mathrm{<span\; class="notranslate">\; R</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; 0.308163</span>}& \mathrm{<span\; class="notranslate">\; 0.227037</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.923907</span>}\\ <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.665353</span>}& \mathrm{<span\; class="notranslate">\; 0.745312</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.038520</span>}\\ <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.679944</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.626832</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 0.380535</span>}\end{array}\right]$

$\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{c}\mathrm{<span\; class="notranslate">\; 47.889089</span>}\\ \mathrm{<span\; class="notranslate">\; 0.686364</span>}\\ \mathrm{<span\; class="notranslate">\; 583.643174</span>}\end{array}\right]$

·Structured light plane equation coefficient

[α b c d] = [-0.013647 -0.013292 -0.190761 19.329410].

The repeatability test errors of the calibrated sensor reconstructing the three-dimensional object coordinates are:

$\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.032</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.0313</span>}\mathrm{<span\; class="notranslate">\; mm</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{{\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.00428</span>}\mathrm{<span\; class="notranslate">\; mm</span>}$

The RMS error of the distance between two points in the measurement space is: E _RMS =0.1496mm.

Claims (1)

1, a kind of scaling method of structured light vision sensor is characterized in that,

(1) set 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;

(2) sensor and target are fixed at a distance of certain distance, should guarantee that the optical plane of laser instrument emission in the sensor and two target faces intersect;

(3) open the power supply of ccd video camera in the sensor, gather the piece image of demarcating target, store in the computing machine;

(4), extract the two-dimensional image coordinate (X on all summits of black square on the target for the target first _i, Y _i) and corresponding three-dimensional article coordinate (x _Wi, y _Wi, z _Wi), store in the computing machine;

(5) open the power supply of laser instrument, make the structured light optical plane of projection and two Plane intersects of target form two-strip structure light striation;

(6) the two-dimensional image coordinate (X of the intersection point at black square edge on extraction structural light strip 5 and the target body _Li, Y _Li), utilize double ratio formula r (A one time _i, B _i, C _i, D _i)=r ' (A _i', B _i', C _i', D _i') calculate as coordinate (X _Li, Y _Li) pairing three-dimensional article coordinate (x _WLi, y _WLi, z _WLi) and storage, r (A in the formula _i, B _i, C _i, D _i) expression A _i, B _i, C _i, D _i4 double ratio, r ' (A _i', B _i', C _i', D _i') expression A _i', B _i', C _i', D _i4 double ratio;

(7) on structural light strip, get arbitrarily a bit, extract it as coordinate (X _Li, Y _Li), the three-dimensional article coordinate (x that tries to achieve in conjunction with double ratio _WLi, y _WLi, z _WLi), utilize dual double ratio invariant r (D ₁, D ₂, D ₃, D ₄)=r ' (D ₁', D ₂', D ₃', D ₄') ask for this point the three-dimensional article coordinate and the storage;

(8) work of repetition (7) can obtain three-dimensional article coordinate and the corresponding two-dimensional image coordinate thereof put on the structured light optical plane of required any amount;

(9) utilize the summit that obtains in (4), to formula

$\left[\begin{array}{c}\mathrm{\&omega;}\&CenterDot;X\\ \mathrm{\&omega;}\&CenterDot;Y\\ \mathrm{\&omega;}\end{array}\right]=\left[\begin{array}{cccc}f& 0& 0& 0\\ 0& f& 0& 0\\ 0& 0& 1& 0\end{array}\right]\left[\begin{array}{cc}R& t\\ 0^T& 1\end{array}\right]\&CenterDot;\left[\begin{array}{c}{x}_w\\ y_w\\ z_w\\ 1\end{array}\right]$ Demarcate, obtain R and t, ω represents a proportionality constant in the formula, and f represents the effective focal length of ccd video camera, and R is 3 * 3 rotation matrix, and t is 3 * 1 translation vector;

(10) point on the structured light optical plane that obtains in utilization (6), (7), (8) is to following formula ax _w+ by _w+ cz _w+ d=0 demarcates, and obtains a, b, c, d, (a, b, c, d) coefficient and the constant term of expression optic plane equations in the formula.

Images