Coaxiality Measurement Method for Docking of Large Cylindrical Frames Based on Monocular Zoom Vision

Abstract

A method for measuring the coaxiality of large barrel-frame docking based on monocular zoom vision, characterized by the following steps: 1) Adjust the zoom camera to the near-focus end, collect images, identify the inner circle features of the front end face of the frame, the inner circle features of the rear end face, and the inner circle features of the front end face of the barrel, and calculate the spatial poses of the inner circles of the front end face of the frame, the inner circles of the rear end face, and the inner circles of the front end face of the barrel; 2) Adjust the zoom camera to the far-focus end, collect images, identify the inner circle features of the rear end face of the barrel, and calculate the spatial pose of the inner circle of the rear end face of the barrel; 3) Fit the axis of the transition frame; 4) Calculate the coaxiality of the front end face and the rear end face of the barrel; 5) After the measurement is completed, judge whether the docking accuracy requirements are met. If so, end the measurement. Otherwise, adjust the pose of the barrel and measure again. The present invention can effectively improve the pose measurement accuracy in the assembly process of aerospace products, and is especially suitable for large barrel-frame docking.

Description

Technical Field

[0001] The present invention relates to a visual measurement technology, especially a two-dimensional image processing or three-dimensional graphics generation technology, specifically a coaxiality measurement method for large barrel frame docking based on monocular zoom vision. Background Art

[0002] In the production process of aerospace products, there are often scenes of large barrel frame docking, which is an important step in the assembly of aerospace products. At present, the domestic large barrel frame docking mainly relies on manual operation by operators. Considering the large weight and size of aerospace products and the high requirements for assembly accuracy and efficiency, the filling method based on manual operation has low efficiency, large filling errors, and requires operators to repeatedly try to install based on past experience.

[0003] In recent years, digital measurement methods mainly based on laser measurement and machine vision have been gradually applied to the assembly and quality inspection processes of aerospace products, improving the product assembly quality and efficiency to a certain extent. However, due to the large size of the barrel sections of aerospace products: the radial dimension is nearly 2 meters, and the axial dimension is about 20 meters, and the high requirements for assembly accuracy: the coaxiality accuracy requirement is ≤3mm, and the working scenarios are variable, which requires higher measurement equipment. Currently, there are few measurement systems for large barrel frame docking applicable to multiple scenarios. Among the existing large barrel frame docking methods, the measurement methods for coaxiality are mostly indirect measurements. By measuring the displacement of the relative reference position on the barrel fixture, the coaxiality of the barrel relative to the axis of the guiding device is judged. This measurement method will introduce the assembly error between the barrel and the fixture. Therefore, the method of directly measuring the coaxiality of the barrel and the guiding device is better. Summary of the Invention

[0004] The present invention aims to provide a coaxiality measurement method for large barrel frame docking based on monocular zoom vision, which can effectively improve the pose measurement accuracy in the assembly process of aerospace products, especially suitable for large barrel frame docking.

[0005] The technical solution of the present invention is as follows:

[0006] A coaxiality measurement method for large barrel frame docking based on monocular zoom vision, comprising the following steps:

[0007] S1: Adjust the zoom camera to the near focus end, collect the images of the barrel and the frame, identify the inner circle features of the front end face, the inner circle features of the rear end face of the frame and the inner circle features of the front end face of the barrel, and calculate the spatial poses of the inner circle of the front end face of the frame, the inner circle of the rear end face of the frame and the inner circle of the front end face of the barrel;

[0008] S2: Adjust the zoom camera to the far focus end, collect the image of the barrel, identify the inner circle feature of the rear end face of the barrel, and calculate the spatial pose of the inner circle of the rear end face of the barrel;

[0009] S3: Fit the axis of the transition frame;

[0010] S4: Calculate the coaxiality of the front end face and the rear end face of the cylinder body;

[0011] S5: After the measurement, determine whether the docking accuracy requirements are met. If they are met, end the measurement; otherwise, adjust the posture of the cylinder body and measure again.

[0012] The identification of the inner circle of the front end face of the frame body, the inner circle of the rear end face, and the inner circle feature of the front end face of the cylinder body in steps S1 and S2, and the identification of the inner circle feature of the rear end face of the cylinder body are carried out according to the following process: Use the Canny algorithm to extract the edge of the circular feature, and then use the improved Zhang Suen algorithm to thin the edge to obtain the edge pixel set E of a single pixel; According to the convexity of the continuous line segments in the pixel set, screen out the arc segments belonging to the edge features of the frame body and the cylinder end face circles to obtain the pixel set E′, and the screening process is as follows:

[0013] S2.1: The pixel set E′ contains several continuous arc segments c1, c2, …, c n , establish a coordinate system with the center of the image as the origin, and divide the arc segments into four groups according to the quadrant where the midpoint position of the arc segment belongs. The arc segments in the first quadrant group The arc segments in the second quadrant group The arc segments in the third quadrant group The arc segments in the fourth quadrant group where n represents that there are n arc segments in the pixel set E′, j represents that there are j arc segments in the first quadrant group, k represents that there are k arc segments in the second quadrant group, l represents that there are l arc segments in the third quadrant group, m represents that there are m arc segments in the fourth quadrant group, and j + k + l + m = n;

[0014] S2.2: Select one arc segment from each group to form a new pixel set E″ for ellipse fitting; Traverse each arc segment in each group to obtain j×k×l×m ellipses, and perform rough screening according to the major and minor axis dimensions of the fitted ellipses, and retain the ellipses that meet the requirements; The i-th combination pixel set that meets the requirements is denoted as E″ i , and the edge pixel set of the fitted ellipse is denoted as e i , where 0 ≤ i ≤ j×k×l×m;

[0015] S2.3: Let the pixel (x i , y i ) be the pixel of the pixel set E″ i , and judge that the number of (x i , y i ) belonging to the pixel set e i is denoted as a i , and the total number of all pixel points of the pixel set E″ i is denoted as b i , The ellipse with the largest value is the ellipse whose edge fits the edge arc segment best, and this ellipse is considered as the end face circle.

[0016] The calculation of the coaxiality reference axis required for fitting the transition box axis and calculating the coaxiality of the front and rear end faces of the cylinder in steps S3 and S4 includes the following steps:

[0017] S3.1: Manually adjust the posture of the cylinder and dock it with the box, and then axially move the cylinder to a fixed distance during measurement;

[0018] S3.2: Collect the image of the front end face of the cylinder at the near focus end and the image of the rear end face of the cylinder at the far focus section, and calculate the center positions of the front end face circle and the rear end face circle of the cylinder;

[0019] S3.3: Repeat the processes of steps S3.1 and S3.2 multiple times to calculate multiple groups of center positions of the end faces;

[0020] S3.4: Calculate the points with the minimum sum of distances to multiple center points of the front end face of the cylinder respectively, and record them as the center reference of the front end face of the cylinder; calculate the points with the minimum sum of distances to multiple center points of the rear end face of the cylinder respectively, and record them as the center reference of the rear end face of the cylinder; specifically: convert to a two-dimensional plane to calculate the reference center of the circle; points P1, P2,..., P 10 That is, 10 center points are obtained, and their coordinates are (x1, y1), (x2, y2),..., (x 10 , y 10 ), and solve the coordinates of the point with the minimum sum of distances to these ten points.

[0021] The calculation of the coaxiality in step S4 includes the following steps. Let the horizontal radial distance from the center of one end face of the cylinder to the reference axis be x m , the vertical radial distance be y m , and the axial distance to the end point of the axis be z m ; the end point of the axis is the intersection of the reference axis and the rear end face of the box; solve for x m , y m and z m according to the spatial position of the axis and the spatial position of the end face circle; establish a box coordinate system: O-XYZ - camera coordinate system; l0 - coaxiality calculation reference axis, (x0, y0, z0) - coordinates of the end point of l0, translate l0 to make the end point coincide with the origin to obtain line l1; l2 - horizontal radial reference perpendicular to line l0 and passing through point (x0, y0, z0), translate l2 to make the end point coincide with the origin to obtain line l3; α - the angle between the projection of l1 on the YOZ plane and Z, β - the angle between l1 and its projection on the YOZ plane, γ - the angle between X after rotating by β (in the same direction) and l3; the box coordinate system takes the end point of the axis as the coordinate origin, l0 as the Z axis, and l2 as the X axis; solve for the center coordinates of the cylinder end face circle in the box coordinate system, that is, [x m y m z m T ​;Set the coordinates of the center of the circular end face of the cylinder in the camera coordinate system to [x y z] T , then

[0022] Description of the Drawings

[0023] Figure 1 is a schematic structural diagram of the system for implementing the present invention;

[0024] Figure 2 is a schematic diagram of the edge feature arc segments obtained after screening the images of each focal length involved in steps S1 and S2 of the present invention;

[0025] Figure 3 is a schematic diagram of the fitting result of the circular face feature involved in steps S1 and S2 of the present invention;

[0026] Figure 4 is a schematic diagram of the solution of the reference center of the circle involved in steps S3 and S4 of the present invention;

[0027] Figure 5 is a schematic diagram of the establishment of the frame coordinate system involved in steps S3 and S4 of the present invention;

[0028] Figure 6 is a schematic diagram of the process of the present invention. Detailed Embodiment

[0029] The present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0030] A method for measuring the coaxiality of large cylinder-frame docking based on monocular zoom vision includes the following steps:

[0031] S1: Adjust the zoom camera to the near focal end, collect images, identify the inner circle of the front end face of the frame, the inner circle of the rear end face, and the inner circle feature of the front end face of the cylinder, and calculate the spatial poses of the inner circle of the front end face of the frame, the inner circle of the rear end face, and the inner circle of the front end face of the cylinder;

[0032] S2: Adjust the zoom camera to the far focal end, collect images, identify the inner circle feature of the rear end face of the cylinder, and calculate the spatial pose of the inner circle of the rear end face of the cylinder;

[0033] S3: Fit the axis of the transition frame;

[0034] S4: Calculate the coaxiality of the front end face and the rear end face of the cylinder;

[0035] S5: After the measurement, determine whether the docking accuracy requirement is met. If it is met, end the measurement; otherwise, adjust the pose of the cylinder and measure again.

[0036] The identification of circular features in steps S1 and S2 is carried out according to the following process: Use the Canny algorithm (Canny J. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8(6): 679-698) to extract the edges of circular features, as Figure 2 shown, and then use the improved Zhang Suen algorithm (Mou Shaomin, Du Haiyang, Su Ping, Zha Xuheng, Chen Guangyi. An improved fast parallel thinning algorithm[J]. Microelectronics & Computer, 2013, 30(01): 53-55+60. DOI: 10.19304 / j.cnki.issn1000-7180.2013.01.014.) to thin the edges to obtain a single-pixel edge pixel set E. According to the convexity of the continuous line segments in the pixel set, the arc segments that may belong to the edge features of the frame and the cylinder end face are screened out to obtain the pixel set E′, and the screening process is as follows:

[0037] S2.1: The pixel set E′ contains several continuous arc segments c1, c2, …, c n , establish a coordinate system with the image center as the origin, and divide the arc segments into four groups according to the quadrant where the midpoint position of the arc segment belongs. The arc segments in the first quadrant group The arc segments in the second quadrant group The arc segments in the third quadrant group The arc segments in the fourth quadrant group where n represents that there are n arc segments in the pixel set E′, j represents that there are j arc segments in the first quadrant group, k represents that there are k arc segments in the second quadrant group, l represents that there are l arc segments in the third quadrant group, m represents that there are m arc segments in the fourth quadrant group, and j + k + l + m = n.

[0038] S2.2: Select one arc segment from each group to form a new pixel set E″ for ellipse fitting. Traverse each arc segment in each group to obtain j×k×l×m ellipses, and perform rough screening according to the major and minor axis dimensions of the fitted ellipses, and retain the ellipses that meet the requirements. The pixel set of the i-th combination that meets the requirements is denoted as E″ i , and the edge pixel set of the fitted ellipse is denoted as e i , where 0 ≤ i ≤ j×k×l×m.

[0039] S2.3: Let the pixel (x i , y i ) be the pixel of the pixel set E″ i , and judge whether (x i , y i ) belongs to the pixel set e iThe number is denoted as a i , the pixel set E″ i The number of all pixel points is denoted as b i , The ellipse with the largest value is the ellipse whose edge fits the edge arc best, and this ellipse is considered as the end face circle, as Figure 3 shown.

[0040] The fitting axis and the calculation of coaxiality in the steps S3 and S4 need to calculate the coaxiality reference axis, which is characterized by including the following steps:

[0041] S3.1: Manually adjust the posture of the cylinder body and dock it with the frame body, and then axially move the cylinder body to a fixed distance during measurement.

[0042] S3.2: Collect the front end face image of the cylinder body at the near focus end and the rear end face image of the cylinder body at the far focus end, and calculate the positions of the front end face circle and the rear end face circle of the cylinder body.

[0043] S3.3: Repeat the processes of steps S3.1 and S3.2 multiple times, and calculate multiple groups of center positions of the end face circles.

[0044] S3.4: Calculate respectively the point with the minimum sum of distances to multiple center points of the front end face, which is denoted as the center reference of the front end face of the cylinder body; the point with the minimum sum of distances to multiple center points of the rear end face, which is denoted as the center reference of the rear end face of the cylinder body. Specifically: Convert to a two-dimensional plane to calculate the reference center of the circle. Points P1, P2, …, P 10 are the 10 obtained center points, and their coordinates are (x1, y1), (x2, y2), …, (x 10 , y 10 ), and solve the coordinates of the point with the minimum sum of distances to these ten points, as Figure 4 shown.

[0045] In the calculation of coaxiality in the step S4, it is necessary to calculate the coaxiality. It is characterized in that the horizontal radial distance from the center of one end face of the cylinder body to the reference axis is x m , the vertical radial distance is y m , and the axial distance to the end point of the axis is z m . The end point of the axis is the intersection point of the reference axis and the rear end face of the frame body. Solve x m , y m and z m according to the spatial positions of the axis and the end face circle. Establish a frame coordinate system, as Figure 5Shown as: O-XYZ - camera coordinate system; l0 - coaxiality calculation reference axis, (x0, y0, z0) - coordinates of the end point of l0, translate l0 to make the end point coincide with the origin to obtain line l1; l2 - horizontal radial reference perpendicular to line l0 and passing through point (x0, y0, z0), translate l2 to make the end point coincide with the origin to obtain line l3; α - the angle between the projection of l1 on the YOZ plane and the Z axis, β - the angle between l1 and its projection on the YOZ plane, γ - the angle between X after rotating by β (in the same direction) and l3. The frame coordinate system takes the axis end point as the coordinate origin, l0 as the Z axis, and l2 as the X axis. Solve for the coordinates of the center of the circular end face of the cylinder in the frame coordinate system, that is, [x m y m z m T . Assume that the coordinates of the center of the circular end face of the cylinder in the camera coordinate system are [x y z] T , then

[0046]

[0047] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention belongs, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

[0048] The parts not involved in the present invention are the same as the prior art or can be implemented by the prior art.​

Claims

1. A method for measuring the coaxiality of large cylinder frame docking based on monocular zoom vision, characterized in that It includes the following steps: The zoom camera is adjusted to the near-focus end to collect images, identify the inner circle features of the front end face of the frame, the inner circle of the rear end face, and the inner circle of the front end face of the cylinder, and calculate the spatial poses of the inner circle of the front end face of the frame, the inner circle of the rear end face, and the inner circle of the front end face of the cylinder; The zoom camera is adjusted to the far-focus end to collect images, identify the inner circle feature of the rear end face of the cylinder, and calculate the spatial pose of the inner circle of the rear end face of the cylinder; S3 Fit the reference axis; S4 Calculate the coaxiality of the front end face and the rear end face of the cylinder; After the measurement is completed, it is judged whether the docking accuracy requirements are met. If so, the measurement ends. Otherwise, the pose of the cylinder is adjusted and the measurement is repeated; The identification of the circular features in steps S1 and S2 is carried out according to the following process: The Canny algorithm is used to extract the edges of the circular features, and then the improved Zhang Suen algorithm is used to thin the edges to obtain a single-pixel edge pixel set E; According to the convexity of the continuous line segments in the pixel set, the arc segments belonging to the circular edge features of the frame and the cylinder end face are screened out to obtain the pixel set E′, and the screening process is as follows: S2.1 The pixel set E' contains several continuous arc segments c1, c2, …, c n , establish a coordinate system with the image center as the origin, and divide the arc segments into four groups according to the quadrant where the midpoint position of the arc segment belongs. The arc segments in the first quadrant group The arc segments in the second quadrant group The arc segments in the third quadrant group The arc segments in the fourth quadrant group where n represents that there are n arc segments in the pixel set E' in total, j represents that there are j arc segments in the first quadrant group, k represents that there are k arc segments in the second quadrant group, l represents that there are l arc segments in the third quadrant group, m represents that there are m arc segments in the fourth quadrant group, and j + k + l + m = n; S2.2 Select an arc segment from each group to form a new pixel set E″ for ellipse fitting; traverse each arc segment in each group to obtain j×k×l×m ellipses, and perform a rough screening based on the major and minor axis dimensions of the fitted ellipses, retaining the ellipses whose dimensions meet the requirements; the pixel set of the i-th combination whose dimensions meet the requirements is denoted as E″ i , and the edge pixel set of the fitted ellipse is denoted as e i , where 0 ≤ i ≤ j×k×l×m; S2.3 Let the pixel (x i , y i ) be a pixel in the pixel set E″ i . Determine the number of (x i , y i ) belonging to the pixel set e i and denote it as a i . Denote the total number of all pixel points in the pixel set E″ i as b i . The ellipse with the largest [[value]] is the ellipse whose edge best fits the edge arc segment, and this ellipse is considered as the end face circle; The fitting of the reference axis in step S3 includes the following steps: S3.1 Manually adjust the pose of the cylinder and dock it with the frame, and then axially move the cylinder to a fixed distance during measurement; S3.2 Collect the image of the front end face of the cylinder at the near-focus end and the image of the rear end face of the cylinder at the far-focus section, and calculate the center positions of the front end face circle and the rear end face circle of the cylinder; S3.3 Repeat the processes of steps S3.1 and S3.2 multiple times to calculate multiple sets of center positions of the end face circles; S3.4 Calculate the points with the minimum sum of distances to multiple center points of the front end face of the cylinder respectively, and record them as the center reference of the front end face of the cylinder; Calculate the points with the minimum sum of distances to multiple center points of the rear end face of the cylinder respectively, and record them as the center reference of the rear end face of the cylinder.

2. The method for measuring the coaxiality of the docking of a large cylindrical frame based on monocular zoom vision according to claim 1, wherein, In step S4, calculating the coaxiality of the front and rear end faces of the cylinder includes the following steps: Let the horizontal radial distance from the center of one end face of the cylinder to the reference axis be x m , the vertical radial distance be y m , and the axial distance from the axis endpoint be z m ; The axis endpoint is the intersection of the reference axis and the rear end face of the frame; Solve for x m , y m and z m according to the spatial position of the reference axis and the spatial position of the circle on this end face of the cylinder; Establish the O-XYZ - camera coordinate system; l0 - the coaxiality calculation reference axis, (x0, y0, z0) - the endpoint coordinates of l0, translate l0 so that the endpoint coincides with the origin to obtain the straight line l1; l2 - the horizontal radial reference perpendicular to the straight line l0 and passing through the point (x0, y0, z0), translate l2 so that the endpoint coincides with the origin to obtain the straight line l3; α - the angle between the projection of l1 on the YOZ plane and the Z axis, β - the angle between l1 and its projection on the YOZ plane, γ - the angle between the X axis after rotating by β and l3; The frame coordinate system takes the axis endpoint as the coordinate origin, l0 as the Z axis, and l2 as the X axis; Solve for the coordinates of the center of the circle on this end face of the cylinder in the frame coordinate system, that is, [x m , y m , z m T ; Let the coordinates of the center of the circle on this end face of the cylinder in the camera coordinate system be [x, y, z] T , then​ 。

Images