Welding Deformation Measurement Method Based on Visual Measurement

Abstract

The present invention relates to a welding deformation measurement method based on vision measurement, comprising the following steps: S1, before welding, using a camera to take a picture of the workpiece that has been spliced but not deformed to obtain a first image; S2, after welding, using the camera to take a picture of the workpiece at the same position to obtain a second image; S3, adopting a boundary feature point extraction algorithm to extract the boundary feature points in the first image; S4, adopting the same boundary feature point extraction algorithm, using the boundary feature points of the first image as the reference points for searching, searching for boundary feature points in the second image with a set search radius, and comparing the pixel distances between the reference points and the searched boundary feature points, and taking the maximum pixel distance as the welding deformation amount. The method of the present invention determines the welding deformation amount by taking pictures of the workpiece before and after welding at the same position, and can be determined by comparative analysis, with high accuracy and high efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of welding, and particularly relates to a welding deformation measurement method based on vision measurement. Background Art

[0002] Welding is an accurate, reliable and low-cost method for connecting materials, and is widely used in fields such as automobiles, ships, aerospace, and mining machinery. It is one of the key manufacturing processes in the manufacturing industry and is irreplaceable. During the welding process, the shape and size changes of the welded workpiece caused by the action of an uneven temperature field are called welding deformation. For all fusion welding methods, there are relatively large residual stresses in the weld and its heat-affected zone. The existence of residual stresses will cause deformation, cracking of the welded component and reduce its bearing capacity. At the same time, stress concentration caused by pits, reinforcement, and undercut exists at the weld toes of the weld. And slag defects and microcracks at the weld toes form the early initiation sources of cracks. In the prior art, after the workpiece is welded, the deformation amount is usually manually inspected. When the deformation is too large, manual correction is required. In short, the accuracy of the manual inspection method for welded workpieces is not high. Summary of the Invention

[0003] The purpose of the present invention is to provide a welding deformation measurement method based on vision measurement, which can improve the measurement accuracy.

[0004] In order to achieve the above invention purpose, the embodiments of the present invention provide the following technical solutions:

[0005] A welding deformation measurement method based on vision measurement includes the following steps:

[0006] S1. Before welding, use a camera to take a picture of the workpiece that has been spliced but not deformed to obtain a first image;

[0007] S2. After welding, use the camera to take a picture of the welded workpiece at the same position to obtain a second image;

[0008] S3. Adopt a boundary feature point extraction algorithm to extract the boundary feature points in the first image;

[0009] S4. Adopt the same boundary feature point extraction algorithm, use the boundary feature points of the first image as the reference points for searching, search for boundary feature points in the second image with a set search radius, and compare the pixel distance between the reference points and the searched boundary feature points, and use the maximum pixel distance as the welding deformation amount.

[0010] In a further optimized solution, in step S1, the workpiece is set in the working space of the welding robot, the welding robot exits the photographing space of the camera, and the camera takes a picture of the workpiece to obtain a first image.

[0011] In a further optimized solution, in step S2, the welding robot welds the workpiece. After welding is completed, the welding robot exits the photographing space of the camera, and the camera takes a photograph of the welded workpiece at the same position to obtain a second image.

[0012] In the above solution, before taking a photograph, first let the welding robot exit the photographing space of the camera and then take a photograph. In this way, the interference of the welding robot in the image can be avoided, so as to more accurately extract the boundary contour, and then improve the measurement accuracy of welding deformation.

[0013] In a further optimized solution, the boundary feature point extraction algorithm uses the Hough line detection algorithm, and the intersection points of two or more Hough lines are used as boundary feature points.

[0014] In a further optimized solution, in step S1, the assembled but undeformed workpieces are assembled with multiple components by using metal glue bonding or spot welding. In this step, each component is assembled and spliced by glue bonding or spot welding to simulate the state after welding. This can not only ensure that the workpiece does not deform during the assembly process, but also ensure the firmness of the connection, and the simulation effect is closer to the state of the workpiece after welding.

[0015] In a further optimized solution, in step S3, the attitude between the camera and the workpiece is obtained from the first image, the boundary feature points to be compared are extracted from the three-dimensional model of the workpiece, transformed into the image coordinate system through the camera calibration matrix, and the extracted boundary feature points to be compared are used as the boundary feature points extracted from the first image.

[0016] The boundary feature points in the first image can be directly extracted from the image or extracted indirectly. In this solution, by extracting the boundary feature points from the three-dimensional model and then converting them into the first image, that is, determining the boundary feature points in the first image by an indirect method, the error generated during image extraction can be reduced, and then the accuracy of the boundary feature points can be improved.

[0017] In a further optimized solution, the step of obtaining the attitude between the camera and the workpiece from the first image includes: placing a standard ball at a specified position of the workpiece before welding, taking a photograph of the workpiece before welding, and obtaining the attitude between the camera and the workpiece by referring to the standard ball.

[0018] The above method is applicable to the situation where it can be ensured that the position is the same before and after welding. The embodiments of the present invention also provide a method applicable to the situation where it cannot be ensured that the position is the same before and after welding. Specifically, a welding deformation measurement method based on vision measurement includes the following steps:

[0019] S1. Before welding, define the workpiece position at this time as P1, and use a camera to take a picture of the assembled but undeformed workpiece to obtain the first image;

[0020] S2. Place the first independent part of the welded workpiece at the same position P1 to obtain the second image;

[0021] S3. Place the welded workpiece, define the workpiece position at this time as P2, use a camera to take a picture of the first independent part of the workpiece before welding to obtain the third image, and use the boundary feature point extraction algorithm to extract the boundary contour points in the second image. Calculate the transformation matrix between the second image and the third image according to the coordinates of the same boundary feature points in the second image and the coordinates in the third image;

[0022] S4. After welding is completed at position P2, use a camera to take a picture of the workpiece to obtain the fourth image;

[0023] S5. Use the same boundary feature point extraction algorithm to extract the boundary feature points in the first image, and transform the boundary feature points to the camera space consistent with position P2 through the transformation matrix;

[0024] S6. Use the same boundary feature point extraction algorithm, use the boundary feature points after coordinate transformation of the first image as the reference points for searching, search for boundary feature points in the fourth image with a set search radius, and compare the pixel distance between the reference points and the searched boundary feature points. Take the maximum pixel distance as the welding deformation amount.

[0025] In a further optimized solution, after the boundary feature point extraction algorithm extracts the boundary feature points in the first image, manual confirmation is performed.

[0026] Compared with the prior art, the present invention has the following beneficial effects: The method of the present invention determines the welding deformation amount by taking pictures of the workpiece before and after welding and through comparative analysis, with high accuracy and higher efficiency compared to manual inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0028] Figure 1 It is a flowchart of the welding deformation measurement method based on vision measurement in Embodiment 1.

[0029] Figure 2It is a flowchart of the welding deformation measurement method based on vision measurement in Embodiment 2. Detailed implementation manners

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The devices in the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0031] Embodiment 1

[0032] Please refer to Figure 1 , a welding deformation measurement method based on vision measurement is provided in this embodiment, including the following steps:

[0033] S1. Before welding, use a camera to take a picture of the workpiece that has been assembled but not deformed to obtain a first image.

[0034] In this step, "assembled but not deformed" means that multiple components that make up a complete workpiece are assembled in a way that does not cause deformation to simulate the state after welding. As a more preferred implementation manner, metal glue bonding or spot welding can be used, which not only does not cause deformation but also ensures the stability of the assembly.

[0035] In this step, when collecting the image, after the workpiece is set in the working space of the welding robot, in order to avoid interference from the welding robot, first let the welding robot exit the camera's photographing space, and then the camera takes a picture of the workpiece to obtain a first image. The first image is the image of the workpiece in the simulated welding state but not deformed and is used as a reference image.

[0036] S2. After welding, use a camera to take a picture of the welded workpiece at the same position to obtain a second image.

[0037] The welding robot welds the workpiece. When the workpiece is welded by multiple components in sequence, that is, component 1 and component 2 are welded first, and then component 3 is welded on component 2, and so on to weld multiple components in sequence.

[0038] In this step, after the welding robot finishes welding the workpiece, the welding robot exits the camera's photographing space, and the camera takes a picture of the welded workpiece at the same position to obtain a second image. The second image is the image of the welded workpiece.

[0039] The same position in this step means that the camera maintains the same pose and takes pictures of the same area of the workpiece, so that the accuracy is higher.

[0040] S3. Extract the boundary feature points in the first image.

[0041] To improve the accuracy of the boundary feature point extraction results, in a more optimized solution, the position of the workpiece is located before collecting the image in step S1. Specifically, a standard ball is placed at a specified position of the workpiece, and then the workpiece before welding is photographed. The pose between the camera and the workpiece is obtained by referring to the standard ball, and then the boundary feature points to be compared are extracted from the three-dimensional model of the workpiece, transformed to the image coordinate system through the camera calibration matrix, and the extracted boundary feature points to be compared are used as the boundary feature points extracted in the first image.

[0042] In this step, the Hough line detection algorithm is used to extract the boundary feature points, and the intersection points of two or more Hough lines are used as the boundary feature points.

[0043] Since the distance between the platform where the workpiece is located and the camera is a known quantity, according to the camera transformation matrix, the boundary feature points in the three-dimensional model of the workpiece determined according to the boundary of the first image are coordinate-transformed to obtain the boundary feature points of the first image.

[0044] It is easy to understand that there is no sequence between steps S3 and S2, and step S3 can be executed after step S1 is completed.

[0045] S4. Using the same boundary feature point extraction algorithm, taking the boundary feature points of the first image as the reference points for searching, searching for boundary feature points in the second image with a set search radius, and comparing the pixel distance between the reference points and the searched boundary feature points, and taking the maximum pixel distance as the welding deformation amount.

[0046] The above measurement method can realize the rapid measurement of the welding deformation amount by comparing the outer contour feature points of the images before and after deformation under the same camera position and pose, and the measurement accuracy is high, and the efficiency is also higher than that of manual detection.

[0047] Embodiment 2

[0048] Please refer to Figure 2 , in this embodiment, another implementation method of the welding deformation measurement method based on vision measurement is provided. This method is more flexible and applicable to more scenarios. Specifically, the method includes the following steps:

[0049] S1. Before welding, define the position of the workpiece at this time as P1, and use the camera to take pictures of the assembled but undeformed workpiece to obtain the first image.

[0050] Similar to step S1 in Embodiment 1, the multiple components can also be assembled by using metal glue bonding to simulate the state after welding, and a reference image, i.e., the first image, is obtained.

[0051] S2. Place the first independent component of the welding workpiece at the same position P1, and use a camera to take a picture of this component to obtain a second image.

[0052] It should be noted here that the camera takes pictures in the same posture as in step S1. Therefore, the visual area of the camera not only includes this component but also the environmental space.

[0053] In addition, the workpiece is formed by welding multiple components together, and each component is independent. When welding, one component is welded to another component. For example, component 1 is welded to component 2, and then to component 3, and so on to complete the welding of all components in sequence. Therefore, the first independent component here refers to component 1, that is, the first component to be placed when preparing for welding.

[0054] The essence of this step is also to obtain a reference image for more accurate coordinate transformation. It is easy to understand that there is no sequence priority between the executions of steps S1 and S2.

[0055] S3. Place the welding workpiece, define the workpiece position at this time as P2, use a camera to take a picture of the first independent component of the workpiece before welding to obtain a third image, and use the boundary feature point extraction algorithm to extract the boundary contour points in the second image. Calculate the transformation matrix between the second image and the third image according to the coordinates of the same boundary feature points in the second image and in the third image.

[0056] In this step, the workpiece is placed arbitrarily, and the workpiece position P2 may be the same as or different from P1. The applicability of this method is stronger, and there is no need to limit that the positions before and after welding must be the same.

[0057] It should be noted that although the workpiece can be placed arbitrarily during welding, in order to reduce the computational difficulty and improve the accuracy, it is preferred that one side of the workpiece positions P1 and P2 coincides, that is, the workpiece is placed to fit a reference positioning edge when placed.

[0058] S4. After the workpiece is welded at position P2, use a camera to take a picture of the welded workpiece to obtain a fourth image.

[0059] S5. Use the same boundary feature point extraction algorithm to extract the boundary feature points in the first image, and transform these boundary feature points to the camera space consistent with position P2 through the transformation matrix.

[0060] S6. Using the same boundary feature point extraction algorithm, taking the boundary feature points after the first image coordinate transformation as the reference points for searching, searching for boundary feature points in the fourth image with a set search radius, and comparing the pixel distances between the reference points and the searched boundary feature points, and taking the maximum pixel distance as the welding deformation amount.

[0061] Similarly, before the camera takes a picture, the welding robot first exits the camera's picture-taking space, and then the camera takes a picture of the workpiece to avoid the occlusion of the welding robot.

[0062] The boundary feature point extraction algorithm can also adopt the Hough line detection algorithm, and the intersection points of more than two Hough lines are used as boundary feature points.

[0063] In order to improve the selection accuracy of boundary feature points, after the boundary feature point extraction algorithm extracts the boundary feature points in the first image, manual confirmation is carried out.

[0064] As described above, only the specific implementation manners of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A welding deformation measurement method based on vision measurement, characterized in that, it includes the following steps: S1. Before welding, use a camera to take a picture of the spliced but undeformed workpiece to obtain a first image; S2. After welding, use the camera to take a picture of the welded workpiece at the same position to obtain a second image; S3. Adopt a boundary feature point extraction algorithm to extract the boundary feature points in the first image; In step S3, obtain the pose between the camera and the workpiece through the first image, extract the boundary feature points to be compared from the three-dimensional model of the workpiece, transform them to the image coordinate system through the camera calibration matrix, and use the extracted boundary feature points to be compared as the boundary feature points extracted in the first image; S4. Adopt the same boundary feature point extraction algorithm, use the boundary feature points of the first image as the reference points for searching, search for boundary feature points in the second image with a set search radius, and compare the pixel distance between the reference points and the searched boundary feature points, and use the maximum pixel distance as the welding deformation amount.

2. The welding deformation measurement method based on vision measurement according to claim 1, characterized in that, in step S1, the workpiece is set in the working space of the welding robot, the welding robot exits the photographing space of the camera, and the camera takes a picture of the workpiece to obtain a first image.

3. The welding deformation measurement method based on vision measurement according to claim 1, characterized in that, in step S2, after the welding robot finishes welding the workpiece, the welding robot exits the photographing space of the camera, and the camera takes a picture of the welded workpiece at the same position to obtain a second image.

4. The welding deformation measurement method based on vision measurement according to claim 1, characterized in that, the boundary feature point extraction algorithm adopts the Hough line detection algorithm, and the intersection points of two or more Hough lines are used as the boundary feature points.

5. A welding deformation measurement method based on vision measurement, characterized in that, it includes the following steps: S1. Before welding, define the workpiece position at this time as P1, and use a camera to take a picture of the spliced but undeformed workpiece to obtain a first image; S2. Place the first independent component of the welding workpiece at the same position P1 to obtain a second image; S3. Place the welding workpiece, define the workpiece position at this time as P2, use a camera to take a picture of the first independent component of the workpiece before welding to obtain a third image, and adopt a boundary feature point extraction algorithm to extract the boundary contour points in the second image, and calculate the transformation matrix between the second image and the third image according to the coordinates of the same boundary feature points in the second image and the coordinates in the third image; S4. After welding is completed at position P2, use a camera to take a picture of the welded workpiece to obtain a fourth image; S5. Adopt the same boundary feature point extraction algorithm to extract the boundary feature points in the first image, and transform the boundary feature points to the camera space consistent with position P2 through the transformation matrix; S6. Using the same boundary feature point extraction algorithm, taking the boundary feature points after the first image coordinate transformation as the reference points for searching, searching for boundary feature points in the fourth image with a set search radius, and comparing the pixel distances between the reference points and the searched boundary feature points, and taking the maximum pixel distance as the welding deformation amount.

6. The welding deformation measurement method based on vision measurement according to claim 5, wherein, in the step S1, the workpiece is set in the working space of the welding robot, the welding robot exits the photographing space of the camera, and the camera takes a photograph of the workpiece to obtain a first image.

7. The welding deformation measurement method based on vision measurement according to claim 5, wherein, in the step S4, after the welding robot finishes welding the workpiece, the welding robot exits the photographing space of the camera, and the camera takes a photograph of the welded workpiece to obtain a second image.

8. The welding deformation measurement method based on vision measurement according to claim 5, wherein, the boundary feature point extraction algorithm adopts the Hough line detection algorithm, and the intersection points of two or more Hough lines are used as boundary feature points.

9. The welding deformation measurement method based on vision measurement according to claim 5, wherein, one side of the workpiece positions P1 and P2 coincides.

10. The welding deformation measurement method based on vision measurement according to claim 5, wherein, after the boundary feature point extraction algorithm extracts the boundary feature points in the first image, manual confirmation is performed.

Images