A multi-hole part pose fast measurement method based on ellipse contour tangent line

Abstract

The application discloses a kind of based on ellipse profile common tangent line's multi-hole part pose fast measurement method.Use a small local area to estimate ellipse tangent line, reduce its tangent line calculation time-consuming.Then it is applied to common tangent line calculation of ellipse profile, improve common tangent line and common tangent point calculation accuracy.Utilize the prior knowledge that part surface has multiple coplanar circular holes to calculate circular hole vanishing point and vanishing line, according to projection geometry, the projection point of circular hole center on image is calculated, the normal vector of circular hole center coordinate and the plane where multiple circular holes are located is obtained by stereo matching, finally, according to the spatial position distribution of circular hole, multi-hole part pose fast estimation is realized.The method proposed in the application reduces the error of ellipse profile tangent line calculation, has higher precision, especially has stronger precision performance under low resolution image, and has faster running speed simultaneously.

Description

Technical Field

[0001] This invention belongs to the field of image detection, and in particular relates to a method for rapid measurement of the pose of porous parts based on the common tangent of an elliptical contour. Background Technology

[0002] Rapid pose estimation of perforated parts using perforated features is a key issue in binocular vision measurement. Existing binocular vision-based methods for perforated part pose estimation typically first measure the pose of each hole, then use the positional relationship between the holes and the part to determine the pose of the perforated part. However, these methods fail to fully utilize the important information that there are multiple coplanar holes on the part surface, resulting in low measurement efficiency. Furthermore, existing methods are significantly affected by image resolution. When the binocular image resolution is low, the number of reference contour points during ellipse detection is limited, leading to increased ellipse detection error. This error propagates to the pose estimation process, increasing the error in perforated hole pose measurement. Therefore, existing methods can only use high-resolution images as input, resulting in high computational load and further reducing efficiency.

[0003] Ellipse tangent estimation error affects the accuracy of circular hole pose measurement by influencing errors in important intermediate quantities such as vanishing point and vanishing line. However, existing methods based on global pixels are sensitive to noise and resolution. Methods based on local vanishing points involve fitting and filtering processes, which are computationally complex, and the error increases with fewer available pixels for fitting and filtering at low resolutions.

[0004] MaWJ et al., building upon monocular vision, calculated the analytical solution for the pose of a circular aperture based on the projection equation of the aperture onto a two-dimensional plane. They then applied geometric constraints to the reconstructed aperture by considering the positional relationship between the two camera coordinate systems, thus eliminating false solutions. XuW et al. estimated the aperture pose twice using monocular vision from two separate images, finding the set with the closest distance and smallest angle between the normal vectors as the true solution. They then constructed an equation to calculate the radius of the aperture. Another approach is to directly use stereo matching with binocular vision for 3D reconstruction of the aperture. LiuY et al. proposed an analytical solution for the pose estimation of a circle. Without any prior knowledge, such as the radius, they used the ellipse equations in the two images and the camera projection model to obtain the quadratic curve equation of the aperture, thus obtaining the spatial pose and radius of the circle. LiuZ et al. established a special coordinate system with the optical center as the origin, calculated a special chord on the aperture in this system, calculated the normal vector of the plane containing the aperture based on the special chord, calculated the vanishing line and the projection of the center onto the left and right images, and reconstructed the center using stereo matching. Peng et al. proposed a target pose measurement method based on maximum outer contour recognition, which can simultaneously achieve both near-range and long-range measurement tasks. Additionally, some researchers have reduced pose measurement errors by improving image quality. For example, Zhang Weiguo used a deep learning-based super-resolution reconstruction model to improve image resolution, thereby enhancing edge extraction quality and achieving high-precision circular hole pose even at lower resolutions. When hardware upgrades are not feasible, this method reduces detection costs and improves accuracy.

[0005] In summary, existing methods for measuring the pose of circular holes based on binocular vision typically utilize two monocular vision models to address the ambiguity in monocular vision measurement, or establish a system of equations based on the projection relationship between the two images and the circular hole to solve for the hole's pose. However, existing methods have the following problems: 1) During the circular hole pose measurement process, due to lens distortion, lighting, and object surface texture, the edge image points of the circular hole contour may deviate from the actual projection points, or even fail to extract the hole's edge. Furthermore, in practical applications, the circular hole is often partially occluded. When these situations occur, the error in ellipse detection increases, and these errors are amplified in subsequent processes, thus affecting the accuracy of pose estimation. 2) Ellipse detection itself introduces errors. Existing methods are highly dependent on the accuracy of ellipse detection; when the ellipse fitting parameters are not precise enough, significant errors occur, resulting in insufficient accuracy and robustness in circular hole pose measurement. 3) Research on pose estimation methods for spatial circles at low resolution is relatively limited. Existing methods often lack sufficient accuracy for low-resolution images. In practical applications, situations frequently arise where image acquisition devices have insufficient resolution or where low-resolution images are required to improve computational speed. Existing methods often fail to meet actual speed requirements. 4) Existing methods fail to utilize prior information about multiple coplanar circular holes on the surface of a part, resulting in high computational complexity and low efficiency, making them difficult to apply to scenarios with high response speed requirements. This limits the application of binocular vision-based circular hole pose measurement in practical engineering. Summary of the Invention

[0006] In order to solve the problems existing in the background art, the purpose of this invention is to provide a rapid measurement method for the pose of multi-hole parts based on the common tangent of the elliptical contour, which has the advantages of speed and high precision.

[0007] This invention uses a small local region to estimate the elliptical tangent, and then applies it to the calculation of the common tangent of the elliptical contour. It uses the prior knowledge that there are multiple coplanar circular holes on the surface of the part to calculate the vanishing point and vanishing line of the holes. It calculates the projection point of the center of the hole on the image according to projective geometry. It obtains the coordinates of the center of the hole and the normal vector of the plane where the multiple holes are located through stereo matching. Finally, it realizes the fast estimation of the pose of the multi-hole part based on the spatial distribution of the holes.

[0008] The specific technical solution adopted in this invention includes the following steps:

[0009] 1) Acquire images of a part with multiple coplanar circular holes using a binocular camera to obtain a target image containing elliptical elements;

[0010] 2) Extract edges from the image and filter out all elliptical contours to obtain all elliptical contours composed of several discrete pixels in the image. Take any point p0 on the elliptical contour and solve for the tangent at point p0 by the tangent estimation method.

[0011] In step 2), the process of finding the tangent at point p0(x0, y0) using the tangent estimation method is as follows:

[0012] Since the equation of the analytic ellipse to which p0 belongs is unknown, the tangent line cannot be calculated analytically. Therefore, this paper utilizes the property that the slope of the line connecting nearby points is approximately equal to the slope of the tangent line at that point, and uses the slope of the line connecting two pixels equidistant from p0's D4 distance to replace the slope of the tangent line to p0:

[0013] 2.1) Each elliptical contour consists of a point set P: P = {p1, p2, ..., p...} n}, take any point p0(x0, y0) on the elliptical contour, and traverse the points in the elliptical contour from near to far on both sides of p0(x0, y0), and find points that are both d away from p0(x0, y0) in two directions, and set them as p1(x1, y1) and p2(x2, y2).

[0014] Where d is a given integer;

[0015] Where any two points p in P i (x i y i ) and p j (x j y j The D4 distance between them is:

[0016] |x i -x j |+|y i -y j |=d ij

[0017] 2.2) Obtain the slope of the line connecting two points p1(x1, y1) and p2(x2, y2):

[0018]

[0019] 2.3) Based on the equation of the straight line Calculate the intercept c:

[0020]

[0021] 3) Using the tangent estimation method in step 2), calculate the vanishing lines of the common tangents of all ellipses;

[0022] Step 3) specifically refers to:

[0023] 3.1) Without loss of generality, assume the contour equations of any two ellipses in the image are as follows:

[0024]

[0025] Among them, A1, B1, C1, D1, E1, F1, A2, B2, C2, D2, E2, and F2 are all coefficients of the ellipse equation;

[0026] Based on the constraint that the centers of both circles are on the same side of the tangent, two external common tangents t are selected. 01 Analytical solutions for t2 and the four common tangent points The two external common tangents t1 and t2 are calculated using the tangent estimation method in step 2). The analytical solutions for the four common tangent points are obtained by solving four sets of solutions to the equation of the ellipse profile. That is, the theoretical value of the common tangent point;

[0027] Where i = 1, 2, represents the i-th circle; j = 1, 2, represents the j-th tangent line;

[0028] 3.2) Traverse the elliptical contour sets P1 and P2 corresponding to the two ellipses, and find the corresponding elliptical contours of the two ellipses respectively. The n points with the shortest distances form a point set G. 11 G 12 G 21 G 22 There are a total of 4n points;

[0029] For G ij For each point in G, ij ={G 11 G 12 G 21 G 22 Using the tangent estimation method in step 2), the tangents of the elliptical contour are obtained, resulting in the set of elliptical contour tangents T. ij ={T 11 T 21 T 12 T 22 The set of tangent lines T of the elliptical contour is obtained using the similarity evaluation method of Hough transform. ij The tangents with the highest similarity are t and t. 11 t 21 t 12 t 22 ;

[0030] 3.3)t 11 and t 21 The corresponding elliptical contour point, i.e., the common tangent point g. 11 and g 21 , t 12 and t 22 The corresponding elliptical contour point, i.e., the common tangent point g. 12 and g 22 g 11 and g 12The equation of the line is l c1 With g 21 and g 22 The equation of the line is l c2 The intersection point is the vanishing point of the projection of the plane containing the two holes onto the image coordinate system;

[0031] Among them, g ij G represents the common tangent point obtained from the set of points P of the elliptical contour, i.e., the actual value of the common tangent point. ij ={g 11 g 12 g 12 g 22};

[0032] 3.4) Traverse all ellipses in the image, calculate a vanishing point for the common external tangent of every two elliptical contours, and the N ellipses in the image can be obtained. There are several vanishing points. The least squares method is used to... The vanishing points are fitted to a vanishing line.

[0033] This article uses T 11 T 21 For example, a similarity evaluation method based on Hough transform is used to evaluate the set of elliptical contour tangents T. ij The similarity between the lines in the graph is given as a score. The tangent line with the highest similarity is found, and its corresponding point of tangency is taken as the common point of tangency, g. 11 and g 21 The linear similarity evaluation method based on Hough transform has high computational speed while ensuring evaluation effectiveness. The specific steps are as follows:

[0034] First, to better represent the case where the tangent slope does not exist, T is represented using polar coordinates. 11 and T 21 The straight line t in 11 and t 21j Next, the Hough transform is used to transform the straight line into a point h in Hough space. 11i :(ρ 11i θ 11i ) and h 21j :(ρ 21j θ 21j ), thus obtaining the point set H of each of the n points in the Hough space. 11 and H 21 Traverse H 11 and H 21 Calculate h 11i With each h 21j The Euclidean distance between them is obtained by n. 2 There are several distances, where the smallest distance corresponds to h. 11i and h 21jFor the nearest point in Hough space, we obtain the two tangent lines t of the elliptical contour with the highest similarity. 11 and t 21 The corresponding elliptical contour point is the common tangent point g. 11 and g 21 Similarly, we obtain the common tangent point g. 12 and g 22 g 11 and g 12 The equation of the line is l c1 With g 21 and g 22 The equation of the line is l c2 The intersection point is the vanishing point of the projection of the plane containing the two holes onto the image coordinate system.

[0035] Since a vanishing point can be calculated from the external common tangent of every two elliptical contours, then for N ellipses in the image, we can obtain... There are several vanishing points. In an ideal binocular vision model, these vanishing points should lie on the vanishing line. However, due to distortion, noise, and the quality of the circular aperture itself, these points are often some distance from the true value of the vanishing line. To reduce the error in the vanishing line calculation process, this paper uses the least squares method to fit the vanishing points to the vanishing line.

[0036] 4) Calculate the projection point of the center of the hole on the image using prior information that multiple holes are located on the same plane, and then realize the three-dimensional reconstruction of the center point of the hole through stereo matching;

[0037] Step 4) specifically involves:

[0038] 4.1) Calculate the projection point of the center of the circular hole in the image using the following method:

[0039] In projective geometry, the transformation formula between coordinates in a single coordinate system in space and the two-dimensional coordinate system of an image is as follows:

[0040] O p =HO (1)

[0041] Where H is a 3×4 projective transformation matrix, and O is the homogeneous coordinate of the circular hole contour points in three-dimensional space. p These are the coordinates of the points on the outline of the circular hole in the two-dimensional coordinate system of the image; the projective transformation of the quadratic curve of the circular hole is as follows:

[0042] e = H -T MH -1 (2)

[0043] Where M represents the 3×3 coefficient matrix of the conic section, M is the coefficient matrix of the circle; e is the coefficient matrix of the circle projected onto the imaging plane to form an ellipse:

[0044]

[0045] Where A, B, C, D, E, and F are the parameters obtained during camera calibration;

[0046] Multiplying equation (1) and equation (2) together, we get:

[0047] eO p =H -T MO

[0048] Right now:

[0049]

[0050] Where R is the radius of the circle;

[0051] The projective transformation of a straight line onto the imaging plane is as follows:

[0052]

[0053] Where l is a three-dimensional vector; The representation of the vanishing line on the imaging plane; l ∞ To represent the vanishing line in three-dimensional space, l ∞ =[0 0 1] T ;

[0054] The comparison shows that eO p and The phases differ by a size factor -R 2 Therefore, the projection point O of the center of the circular hole p It can be obtained by multiplying the inverse of the elliptic coefficient matrix by the vanishing line of the plane containing the circular hole:

[0055]

[0056] 4.2) For each circular hole on the part, set the elliptic coefficient matrix of the hole in the left and right images as e and e′ respectively, and obtain the equation of the vanishing line of the plane where the hole is located. and Therefore, according to step 4.1), the projected image points of the center of the circular hole are respectively and

[0057] 4.3) The optimal triangulation method is used to achieve three-dimensional reconstruction of the center of the circular hole, that is, through the two projection points O of the center of the circle. p and O′ p The three-dimensional coordinates of the center of the circular hole are obtained.

[0058] This method employs a non-iterative approach based on the projection points O of the two center points. p and O′ pFinding the position of the center of the circular hole in three-dimensional space can guarantee that the solution obtained is optimal with low computational complexity.

[0059] 5) Perform pose estimation based on the spatial distribution of multiple holes in the porous part.

[0060] Step 5) specifically involves:

[0061] Based on the three-dimensional coordinates of the center of the circular hole obtained in step 4), and according to the coplanar constraint that the planes containing the multiple circular holes are the same plane, the three-dimensional coordinates of the center points of the multiple circular holes are fitted with the least squares plane to calculate the normal vector of the plane containing the circular holes, that is, to obtain the representation of the plane containing the circular holes in three-dimensional space. Then, a part coordinate system is established with this plane, and the pose of the part in three-dimensional space is represented by the established part coordinate system.

[0062] This invention addresses the problem of pose estimation for porous parts based on binocular vision measurement, proposing a fast pose estimation method for porous parts based on the common tangent of an elliptical profile.

[0063] This invention utilizes local pixels to calculate the tangent of the elliptical contour, reducing the calculation error of the elliptical contour tangent and improving the calculation speed. This method has stronger accuracy in low-resolution images.

[0064] This invention improves the calculation accuracy of common tangents and common tangent points, and realizes pose estimation for multi-hole parts.

[0065] The beneficial effects of this invention are:

[0066] 1) The method proposed in this invention reduces the calculation error of the tangent of the elliptical contour and has high accuracy.

[0067] 2) This invention has stronger accuracy performance in low-resolution images, while also having a faster running speed. Attached Figure Description

[0068] Figure 1 It refers to the positional relationship between multiple circular holes on a part.

[0069] Figure 2 The results of the method of the present invention in analyzing ellipse, pixel discrete ellipse contour, and actual circular hole edge are shown in (a), (b), and (c), which represent continuous ellipse, discrete ellipse, and actual circular hole edge, respectively.

[0070] Figure 3 These are the vanishing points and vanishing lines corresponding to the plane containing the multiple circular holes.

[0071] Figure 4 It involves establishing the coordinate system of the part.

[0072] Figure 5 Comparison of experimental results on the change of average error relative to the semi-major axis Detailed Implementation

[0073] The present invention will be further described below with reference to the accompanying drawings and examples.

[0074] The embodiments and implementation process of the present invention are as follows:

[0075] 1) Acquire target images containing elliptical elements, specifically by acquiring images of parts with multiple coplanar circular holes.

[0076] 2) After edge extraction and filtering of the elliptical contours in the image, all elliptical contours composed of discrete pixels are obtained. For each elliptical contour, i.e., a point set P:

[0077] P = {p1, p2, ..., p} n}

[0078] Suppose we want to find the tangent line at a point p0(x0, y0) on an elliptical contour. Since the analytical equation of the ellipse to which p0 belongs is unknown, the tangent line cannot be calculated analytically. Therefore, this paper utilizes the property that the slope of the line connecting nearby points is approximately equal to the slope of the tangent line at that point, and substitutes the slope of the tangent line at p0 with the slope of the line connecting two pixels equidistant from p0 by D4 distance. For any point p in P... i (x i y i Distance p j (x j y j The D4 distance of ) is:

[0079] |x i -x j |+|y i -y j |=d ij

[0080] Given an integer d, the specific steps to find the tangent line at p0(x0, y0) are as follows:

[0081] 2.1) Traverse the points in the elliptical contour sequentially from near to far on both sides of p0(x0, y0). Find the points that are both d away from p0(x0, y0) in both directions. Considering the continuity of the edge, in a small local region, there are only two points with a distance of d on each side, denoted as p1(x1, y1) and p2(x2, y2).

[0082] 2.2) Find the slope of the line connecting two points p1(x1, y1) and p2(x2, y2).

[0083]

[0084] 2.3) Find the intercept c: The equation of the straight line is given by the following formula:

[0085]

[0086] The above algorithm is applied to the calculation results of ellipse and pixel ellipse contours, as shown below. Figure 2 As shown.

[0087] 3) such as Figure 3 As shown, the vanishing line of the common tangent of the elliptical profile is calculated using the tangent estimation method.

[0088] Without loss of generality, for any two ellipses in the image, assume their general equations are as follows:

[0089]

[0090] Using the above analytical equations, we find four solutions to the system of two quadratic equations in two variables. Then, based on the constraint that the centers of the two circles are on the same side of the tangent, we select two external common tangents t1 and t2 and four common tangent points to find analytical solutions. Where i = 1, 2 represents the i-th circle, and j = 1, 2 represents the j-th tangent line. Then, iterate through the elliptical contours P1 and P2 corresponding to the two ellipses. Find the... There are n points with the shortest distance, totaling 4n points. These 4n points form a point set G. 11 G 12 G 21 G 22 For G ij For each point in the equation, find the tangent line to the ellipse contour using the method in Section 4.2, and obtain the set of ellipse contour tangent lines T. ij Among them, T 11 and T 21 Corresponding tangent T 12 and T 22 Corresponding tangent

[0091] This article uses T 11 T 21 For example, a similarity evaluation method based on Hough transform is used to score the similarity of lines in the pair, and the tangent line with the highest similarity is found, with its corresponding tangent point being taken as the common tangent point g. 11 and g 21 The linear similarity evaluation method based on Hough transform has high computational speed while ensuring evaluation effectiveness. The specific steps are as follows:

[0092] First, to better represent the case where the tangent slope does not exist, T is represented using polar coordinates. 11 and T 21 The straight line t in 11 and t 21jNext, the Hough transform is used to transform the straight line into a point h in Hough space. 11i :(ρ 11i θ 11i ) and h 21j :(ρ 21j θ 21j ), thus obtaining the point set H of each of the n points in the Hough space. 11 and H 21 Traverse H 11 and H 21 Calculate h 11i The Euclidean distance between each h21j is obtained to get n. 2 There are several distances, where the smallest distance corresponds to h. 11i and h 21j For the nearest point in Hough space, we obtain the two tangent lines t of the elliptical contour with the highest similarity. 11 and t 21 The corresponding elliptical contour point is the common tangent point g. 11 and g 21 Similarly, we obtain the common tangent point g. 12 and g 22 g 11 and g 12 The equation of the line is l c1 With g 21 and g 22 The equation of the line is l c2 The intersection point is the vanishing point of the projection of the plane containing the two holes onto the image coordinate system.

[0093] Since a vanishing point can be calculated from the external common tangent of every two elliptical contours, then for N ellipses in the image, we can obtain... There are several vanishing points. In an ideal binocular vision model, these vanishing points should lie on the vanishing line. However, due to distortion, noise, and the quality of the circular aperture itself, these points are often some distance from the true value of the vanishing line. To reduce the error in the vanishing line calculation process, this paper uses the least squares method to fit the vanishing points to the vanishing line.

[0094] 4) Utilize prior information that multiple circular holes lie on the same plane to calculate the projection point of the hole center onto the image, and then achieve 3D reconstruction of the hole center point through stereo matching. In projective geometry, the conversion formula between coordinates in a spatial coordinate system and the image's two-dimensional coordinate system is as follows:

[0095] O p =HO

[0096] Where H is a 3×4 projective transformation matrix, and O is the homogeneous coordinate of the points on the circular hole contour in three-dimensional space. The projective transformation of the quadratic curve including the circular hole is as follows:

[0097] e = H -T MH -1

[0098] in

[0099] Where M represents the 3×3 coefficient matrix of the conic section. M is the coefficient matrix of the circle. e is the coefficient matrix of the circle projected onto the imaging plane to form an ellipse. Multiplying the above equations together, we get:

[0100] eO p =H -T MO

[0101] In the space plane π, a circle with center (X0, Y0) and radius R is represented as follows:

[0102] XJX T =0 of which

[0103] get:

[0104]

[0105] When a straight line is projected onto the imaging plane, its projective transformation is as follows:

[0106]

[0107] Where l is a three-dimensional vector. Since the vanishing line is a straight line at infinity on the plane, its vector representation is l. ∞ =[0 0 1] T The comparison shows that eO p With l p The phases differ by a size factor -R 2 Therefore, the projection point of the center of the circular hole can be obtained by multiplying the inverse of the elliptic coefficient matrix with the vanishing line of the plane containing the circular hole:

[0108]

[0109] Thus, given the elliptic coefficient matrix and the vanishing line equation coefficients, this paper can calculate the projection points of the centers of all circular holes in the image using the above formula. For each circular hole on the part, assume that its elliptic coefficient matrices in the left and right images are e and e′, respectively, and the vanishing line equations of the plane containing the hole are respectively... and Then its projected image points are respectively and Finally, the optimal triangulation method proposed by Hartley et al. can be used to achieve the three-dimensional reconstruction of the center of the circular hole. This method employs a non-iterative approach based on the projection points O of the two center points. p and O′ pFinding the position of the center of the circular hole in three-dimensional space can guarantee that the solution obtained is optimal with low computational complexity.

[0110] 5) Pose estimation for multi-hole parts based on the spatial distribution of circular holes. After obtaining the coordinates of the hole centers, firstly, based on the coplanar constraint that the planes containing multiple holes are the same plane, least-squares plane fitting is used to calculate the normal vector of the plane containing the holes in the three-dimensional coordinates of the multiple hole centers. Then, a part coordinate system is established. This three-dimensional coordinate system remains relatively stationary with the part and represents the part's pose in three-dimensional space. The origin of the coordinate system represents the part's position in three-dimensional space. It should facilitate subsequent gripping operations on the part while ensuring its fixed geometric relationship with the multiple holes. The three axes of the rectangular coordinate system represent the part's orientation in three-dimensional space. The directions should, as far as possible, maintain a special positional relationship with the axes and planes within the part, such as parallel or perpendicular, to facilitate subsequent calculations and operations. Figure 4 In the part shown, the origin of the coordinate system is the center point of all the circular holes, obtained by averaging the coordinates of all the hole centers. The Z-axis of the coordinate system is parallel to the plane normal vector, and its direction vector is the plane normal vector. The X-axis of the coordinate system is the vector obtained by connecting it to circular holes 3 and 4 in the figure. The Y-axis is obtained by the cross product of the Z-axis and the X-axis.

[0111] 6) Experimental measurement results.

[0112] 6.1) Ellipse Profile Tangent Estimation Experiment. To examine the effectiveness of the tangent estimation method proposed in this chapter, a tangent estimation experiment was conducted and compared with existing methods. Considering the size of the ellipse in the actual binocular image, five sets of ellipses were generated, with different semi-major axes 'a' for each set. Tangent estimation was performed using the method in this chapter, the Yumnam method, and the IPF method. The Yumnam method treats the tangent slope as a function of a maximal straight line segment, estimating the tangent by finding this maximal straight line segment. The IPF method fits a continuous curve based on several nearby points, replacing the slope of the discrete curve with the slope of the continuous curve. Each set contains five ellipses, and each ellipse has ten points, resulting in 50 tangent slope error values. The average of these 50 errors was calculated to obtain the mean absolute error, which varies with the semi-major axis 'a' as follows: Figure 5 As shown.

[0113] 6.2) Circular Hole Pose Measurement Experiment. The center of the six simulated circular holes in the image was located using the method described in this paper as the estimated center location value. The distance between the center and the center of the four corner points of the white square was taken as the center location error. The Xu method and Liu method were selected for comparison, and the measurement results are shown in Table 1.

[0114] Table 1 Comparison of Experimental Results for Center Positioning

[0115]

[0116] 6.3) Runtime Experiment. In addition, to examine the computational speed of the proposed method, comparative experiments were conducted on a laptop with an Intel Core i5-10210 processor and 16GB of RAM. All pose estimation methods were implemented in C++11, with an image resolution of 2448 pixels × 2048 pixels. Ten pose estimations were performed, and the average measurement time is shown in Table 2.

[0117] Table 2 Comparison of Experimental Results for Measurement Time

[0118]

[0119] Based on the above experimental results, it can be seen that the method proposed in this invention reduces the calculation error of the elliptical contour tangent and has high accuracy, especially in low-resolution images, while also having a fast running speed.

[0120] The above embodiments should not be considered as limitations on the present invention, but any improvements made based on the spirit of the present invention should be within the protection scope of the present invention.

Claims

1. A method for rapid pose measurement of multi-hole parts based on the common tangent of an elliptical profile, characterized in that, Includes the following steps: 1) Acquire images of a part with multiple coplanar circular holes using a binocular camera to obtain a target image containing elliptical elements; 2) Perform edge extraction on the image and filter out all elliptical contours to obtain all elliptical contours composed of a number of discrete pixels in the image. Then, arbitrarily select a point on one of the elliptical contours. Solving for points using the tangent estimation method Tangent at the point; 3) Using the tangent estimation method in step 2), calculate the vanishing lines of the common tangents of all ellipses; 4) Calculate the projection point of the center of the hole on the image using prior information that multiple holes are located on the same plane, and then realize the three-dimensional reconstruction of the center point of the hole through stereo matching; 5) Perform pose estimation based on the spatial distribution of multiple circular holes in the porous part; In step 2), the point is solved using the tangent estimation method. The process of finding the tangent is as follows: Adopted and The slope of the line connecting two pixels equidistant from each other in D4 is used to replace tangent slope: 2.1) Each elliptical contour consists of a set of points. composition: Choose any point on the elliptical outline ,exist Traverse the points in the elliptical outline sequentially from near to far on both sides, and find the points along both directions that correspond to the points in the ellipse. The distances are all The points are divided into and ; in, Given an integer; 2.2) Obtain two points and The slope of the line connecting the two lines: 2.3) Calculate the intercept : Step 3) specifically refers to: 3.1) Suppose the contour equations of any two ellipses in the image are as follows: in, , All are coefficients of the elliptic equation; Based on the constraint that the centers of both circles are on the same side of the tangent, two external common tangents are selected. and and analytical solutions for the four common tangent points Two external tangents and The tangent estimation method in step 2) is used to calculate the ellipse profile equation, and the analytical solutions for the four common tangent points are obtained by solving four sets of solutions. , { , , , }, that is, the theoretical value of the common tangent point; in, , indicating the first A circle; , indicating the first Tangent line; 3.2) Traverse the set of elliptical outlines corresponding to the two ellipses. , Find the corresponding , , , Shortest distance Each of the points forms a point set. , , , ,common One point; right Each point in, , , , Using the tangent estimation method in step 2), the tangents of the elliptical contour are obtained, resulting in a set of elliptical contour tangents. The set of tangent lines for the elliptical contour is obtained using the similarity evaluation method of Hough transform. The tangents with the highest similarity are respectively , , , ; 3.3) and The corresponding elliptical contour points are the common tangent points. and , and The corresponding elliptical contour points are the common tangent points. and ; and The equation of the line and and The equation of the line The intersection point is the vanishing point of the projection of the plane containing the two holes onto the image coordinate system; in, Represents the set of points of an ellipse contour The obtained common tangent point, i.e., the actual value of the common tangent point, ; 3.4) Traverse all ellipses in the image, and calculate a vanishing point for the common external tangent of every two elliptical contours. An ellipse can be obtained There are several vanishing points. The least squares method is used to... The vanishing points are fitted to a vanishing line. .

2. The method for rapid pose measurement of multi-hole parts based on the common tangent of an elliptical contour according to claim 1, characterized in that, Step 4) specifically involves: 4.1) Calculate the projection point of the center of the circular hole in the image using the following method: In projective geometry, the transformation formula between coordinates in a single coordinate system in space and the two-dimensional coordinate system of an image is as follows: （1） in, It is a 3×4 projective transformation matrix. These are the homogeneous coordinates of the points on the circular hole's outline in three-dimensional space. These are the coordinates of the points on the outline of the circular hole in the two-dimensional coordinate system of the image; the projective transformation of the quadratic curve of the circular hole is as follows: （2） in, The 3×3 coefficient matrix representing a conic section is the same as the coefficient matrix of a circle. It is the coefficient matrix of the circle projected onto the imaging plane to form an ellipse: Where A, B, C, D, E, and F are the parameters obtained during camera calibration; Multiplying equation (1) and equation (2) together, we get: Right now: in, Let be the radius of the circle; The projective transformation of a straight line onto the imaging plane is as follows: in, It is a three-dimensional vector; The representation of the vanishing line on the imaging plane; This represents the vanishing line in three-dimensional space. ; The projection point of the center of the circular hole is obtained by multiplying the inverse of the elliptic coefficient matrix with the vanishing line of the plane containing the circular hole. : 4.2) For each circular hole on the part, set the elliptic coefficient matrix of the circular hole in the left and right images respectively as follows: and And obtain the equations of the vanishing lines of the plane containing the circular holes. and Thus, according to step 4.1), the projected image points of the center of the circular hole are respectively... and ; 4.3) The optimal triangulation method is used to achieve three-dimensional reconstruction of the center of the circular hole, that is, through two projection points of the center of the circle. and The three-dimensional coordinates of the center of the circular hole are obtained.

3. The method for rapid pose measurement of multi-hole parts based on the common tangent of an elliptical contour according to claim 1, characterized in that, Step 5) specifically involves: Based on the three-dimensional coordinates of the center of the circular hole obtained in step 4), and according to the coplanar constraint that the planes containing the multiple circular holes are the same plane, the three-dimensional coordinates of the center points of the multiple circular holes are fitted with the least squares plane to calculate the normal vector of the plane containing the circular holes, that is, to obtain the representation of the plane containing the circular holes in three-dimensional space. Then, a part coordinate system is established with this plane, and the pose of the part in three-dimensional space is represented by the established part coordinate system.

Images