3D model reconstruction method for welds

Abstract

Provided is a 3D model reconstruction method for a weld, by converting multiple sets of two-dimensional scanning results into three-dimensional point cloud data of the weld in a welding gun coordinate system, and then extracting a spatial plane model according to the scanning weld data characteristics, and reconstructing a spatial data set matching the welding plane, the measurement error caused by the inconsistency between the measurement plane and the welding plane is solved, so as to obtain accurate weld appearance characteristics, solve the problem that the traditional method cannot truly reflect the specific change characteristics of the weld, improve the weld model precision and quality, accurately position the appearance characteristics and position of the weld, and thus improve the welding quality and precision.

Description

Technical Field

[0001] This invention belongs to the field of weld seam modeling technology, specifically relating to a 3D model reconstruction method for weld seams. Background Technology

[0002] In the process of weld scanning, the scanning results in the laser coordinate system are often used directly to calculate the position of the weld by its relative position to the welding torch tip. The above method maps the information of the three-dimensional weld onto the two-dimensional scanning plane. When the laser scanning plane and the welding torch plane are not parallel, there will be an angular deviation in the result, which will cause the calculation result to fail to reflect the specific changes of the weld. Moreover, the three-dimensional weld model often produces problems such as holes or blurry details, making it impossible to accurately locate the weld position. Therefore, it is necessary to improve the method to address the above problems. Summary of the Invention

[0003] The technical problem solved by this invention is to provide a 3D model reconstruction method for welds. By converting multiple sets of two-dimensional scanning results into three-dimensional point cloud data of the weld in the welding gun coordinate system, and then extracting a spatial plane model based on the characteristics of the scanned weld data, a spatial dataset matching the welding plane is reconstructed. This solves the measurement error caused by the inconsistency between the measurement plane and the welding plane, thereby obtaining accurate weld morphology features. It also solves the problem that traditional methods cannot truly reflect the specific changes in the weld, improves the accuracy and quality of the weld model, and accurately locates the morphology features and position of the weld, thereby improving welding quality and accuracy.

[0004] The technical solution adopted in this invention is a 3D model reconstruction method for welds, comprising the following steps:

[0005] 1) Use a laser profilometer to scan the weld and obtain the contour data of the weld and its surrounding area;

[0006] 2) Based on the collected two-dimensional weld contour data information on the x-axis and z-axis, and the y-axis data constructed according to the scanning speed, three-dimensional point cloud data of the weld is formed.

[0007] 3) Radius filtering is used to remove outliers from the constructed 3D point cloud data of the weld seam;

[0008] 4) After smoothing the 3D point cloud data using the moving least squares method, a flawless and smooth surface model is established.

[0009] 5) Extracting spatial plane equations from 3D point cloud data: Randomly select three points in the initial point cloud and calculate their corresponding plane equations Ax + By + Cz + D = 0. Then, according to the formula d... i =|Ax i +By i +Czi +D| Calculate the algebraic distance between all points and the equation of the plane, and select a threshold d. threshold If d i ≤d threshold If the number of inliers is positive, the point is considered an in-model sample point; otherwise, it is an out-of-model sample point. Record the current number of inliers. Using the obtained plane equation, calculate the ratio of the number of inliers to the total number of sample points. The plane equation with the highest ratio is the best-fit parameter; that is, the best-fit parameter is the plane equation corresponding to the plane with the most inliers. Next, calculate the error rate Δ at the end of each iteration, using the following formula:

[0010]

[0011] In the above formula, error mean The error is the average of the distance differences between all non-interior points and the corresponding interior points in the plane equations. min The minimum difference in distances from all non-interior points to the plane equations corresponding to interior points is called error. max It is the maximum value among all the distance differences from non-interior points to the plane equations corresponding to interior points;

[0012] Then, based on the determined number of interior points N inliers The total number of samples N in the three-dimensional space is used to calculate the iteration termination evaluation factor δ. The formula for calculating the iteration termination evaluation factor δ is as follows:

[0013]

[0014] In the above formula, ω0 is the set proportional coefficient of 0.7, and ω1 is the set proportional coefficient of 0.3. If the evaluation factor δ does not fall within the interval [0,1] after the iteration ends, three points are randomly selected in the initial point cloud, and this step is repeated for the next iteration. If the evaluation factor δ falls within the interval [0,1] after the iteration ends, the iteration stops. After the iteration ends, the optimal model parameters are the extracted spatial plane equation.

[0015] 6) Based on the type of weld on the target workpiece, determine the number of planes F of that type of weld. Following step 5), determine the first spatial plane equation. Separate the interior points on the determined first spatial plane equation from the three-dimensional point cloud data. Repeat step 5) on the remaining three-dimensional point cloud data to determine the same number of spatial plane equations as the number of planes F.

[0016] 7) Based on the obtained F spatial plane equations, filter out the point cloud set within all spatial plane equations from the original laser profilometer scan data, and reconstruct the weld model;

[0017] 8) From the reconstructed 3D weld model, determine the plane normal vectors of F spatial plane equations. Based on the input weld angle α, calculate the angle β between the plane normal vectors of the F spatial plane equations according to the direction of travel. Take the two normal vectors when Δt=|α-β| is the minimum, and determine the spatial plane equations corresponding to these normal vectors, which are the two weld side surfaces where the weld is located.

[0018] 9) Obtain the equation of the intersecting line in space by extending the two planes in space. From the starting point of the scan to the ending point of the scan, obtain the line segment on the line equation, which is the weld position.

[0019] Advantages of this invention compared to existing technologies:

[0020] 1. This technical solution converts multiple sets of two-dimensional scanning results into three-dimensional point cloud data of the weld in the welding gun coordinate system. Then, based on the characteristics of the scanned weld data, it extracts a spatial plane model and reconstructs a spatial dataset that matches the welding plane. This solves the measurement error caused by the inconsistency between the measurement plane and the welding plane, thereby obtaining accurate weld morphology features. It solves the problem that traditional methods cannot truly reflect the specific changes in the weld, and improves the accuracy and quality of the weld model.

[0021] 2. The process of determining the weld position in this technical solution is easy to operate, accurately locates the morphological features and position of the weld, and provides a precise and reliable position reference for unmanned operation on the welding site, thereby improving welding quality and accuracy.

[0022] 3. This technical solution differs from the previous method of calculating weld morphology using relative measurement values. Relative measurement values ​​are limited to the results of a single measurement and lack the spatial relationship between the data from previous and subsequent scans. In contrast, absolute measurement values ​​can truly reflect the spatial relationship between the data, thereby calculating more accurate morphological features.

[0023] 4. This technical solution, by employing spatial datasets and spatial data relationships, can reduce the impact of noise signals on the dataset and has strong noise resistance. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0025] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0026] The method for reconstructing a 3D model of a weld includes the following steps:

[0027] 1) Use a laser profilometer to scan the weld and obtain the contour data of the weld and its surrounding area;

[0028] 2) Based on the collected two-dimensional weld contour data along the x and z axes, and the y-axis data constructed according to the scanning speed, three-dimensional point cloud data of the weld is formed; by adopting the world coordinate system, the absolute spatial position is constructed, and the 3D structure of the weld is digitally and realistically output.

[0029] 3) Radius filtering is used to remove outliers from the constructed 3D point cloud data of the weld seam;

[0030] 4) After smoothing the 3D point cloud data using the moving least squares method, a flawless and smooth surface model is established.

[0031] 5) Extracting spatial plane equations from 3D point cloud data: Randomly select three points in the initial point cloud and calculate their corresponding plane equations Ax + By + Cz + D = 0. Then, according to the formula d... i =|Ax i +By i +Cz i +D| Calculate the algebraic distance between all points and the equation of the plane, and select a threshold d. threshold If d i ≤d threshold If the number of inliers is positive, the point is considered an in-model sample point; otherwise, it is an out-of-model sample point. Record the current number of inliers. Using the obtained plane equation, calculate the ratio of the number of inliers to the total number of sample points. The plane equation with the highest ratio is the best-fit parameter; that is, the best-fit parameter is the plane equation corresponding to the plane with the most inliers. Next, calculate the error rate Δ at the end of each iteration, using the following formula:

[0032]

[0033] In the above formula, error mean The error is the average of the distance differences between all non-interior points and the corresponding interior points in the plane equations.min The minimum difference in distances from all non-interior points to the plane equations corresponding to interior points is called error. max It is the maximum value among all the distance differences from non-interior points to the plane equations corresponding to interior points;

[0034] Then, based on the determined number of interior points N inliers The total number of samples N in the three-dimensional space is used to calculate the iteration termination evaluation factor δ. The formula for calculating the iteration termination evaluation factor δ is as follows:

[0035]

[0036] In the above formula, ω0 is the set proportionality coefficient of 0.7, and ω1 is the set proportionality coefficient of 0.3. If the evaluation factor δ does not fall within the interval [0,1] after the iteration ends, three points are randomly selected in the initial point cloud, and this step is repeated for the next iteration. If the evaluation factor δ falls within the interval [0,1] after the iteration ends, the iteration stops. After the iteration ends, the optimal model parameters are the extracted spatial plane equations. This step can minimize the error in weld parameter calculation caused by misidentification of the plane or misalignment of the identified plane.

[0037] 6) Based on the type of weld on the target workpiece, determine the number of planes F of that type of weld. Following step 5), determine the first spatial plane equation. Separate the interior points on the determined first spatial plane equation from the three-dimensional point cloud data. Repeat step 5) on the remaining three-dimensional point cloud data to determine the same number of spatial plane equations as the number of planes F.

[0038] 7) Based on the obtained F spatial plane equations, filter out the point cloud set within all spatial plane equations from the original laser profilometer scan data, and reconstruct the weld model;

[0039] 8) From the reconstructed 3D weld model, determine the plane normal vectors of F spatial plane equations. Based on the input weld angle α, calculate the angle β between the plane normal vectors of the F spatial plane equations according to the direction of travel. Take the two normal vectors when Δt = |α - β| is minimized, and determine the spatial plane equations corresponding to these normal vectors, which are the two weld side surfaces where the weld is located. This step directly and accurately identifies the weld bead that constitutes the weld, improves the accuracy of the measurement and calculation results of the actual position of the weld, changes the previous method of locating the weld by a fixed offset, and increases the ability to adapt to the dimensional errors of the weld caused by assembly and processing.

[0040] 9) Obtain the equation of the intersecting line in space by extending the two planes in space. From the starting point of the scan to the ending point of the scan, obtain the line segment on the line equation, which is the weld position.

[0041] This invention converts multiple sets of two-dimensional scanning results into three-dimensional point cloud data of the weld seam in the welding torch coordinate system. Then, based on the characteristics of the scanned weld seam data, it extracts a spatial plane model and reconstructs a spatial dataset that matches the welding plane. This solves the measurement error caused by the inconsistency between the measurement plane and the welding plane, thereby obtaining accurate weld seam morphology features. It also solves the problem that traditional methods cannot truly reflect the specific changes in the weld seam, improving the accuracy and quality of the weld seam model. The process of determining the weld seam position is easy to operate, accurately locating the morphology features and position of the weld seam, providing a precise and reliable position benchmark for unmanned operations on the welding site, thereby improving welding quality and accuracy.

[0042] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

[0043] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for reconstructing a 3D model of a weld, characterized in that... Includes the following steps: 1) Use a laser profilometer to scan the weld and obtain the contour data of the weld and its surrounding area; 2) Based on the collected two-dimensional weld contour data information on the x-axis and z-axis, and the y-axis data constructed according to the scanning speed, three-dimensional point cloud data of the weld is formed. 3) Radius filtering is used to remove outliers from the constructed 3D point cloud data of the weld seam; 4) After smoothing the 3D point cloud data using the moving least squares method, a flawless and smooth surface model is established. 5) Extracting spatial plane equations from 3D point cloud data: Randomly select three points in the initial point cloud and calculate their corresponding plane equations Ax + By + Cz + D = 0. Then, according to the formula d... i =|Ax i +By i +Cz i +D| Calculate the algebraic distance between all points and the equation of the plane, and select a threshold d. threshold If d i ≤d threshold If the number of inliers is positive, the point is considered an in-model sample point; otherwise, it is an out-of-model sample point. Record the current number of inliers. Using the obtained plane equation, calculate the ratio of the number of inliers to the total number of sample points. The plane equation with the highest ratio is the best-fit parameter; that is, the best-fit parameter is the plane equation corresponding to the plane with the most inliers. Next, calculate the error rate Δ at the end of each iteration, using the following formula: In the above formula, error mean The error is the average of the distance differences between all non-interior points and the corresponding interior points in the plane equations. min The minimum difference in distances from all non-interior points to the plane equations corresponding to interior points is called error. max It is the maximum value among all the distance differences from non-interior points to the plane equations corresponding to interior points; Then, based on the determined number of interior points N inliers The total number of samples N in the three-dimensional space is used to calculate the iteration termination evaluation factor δ. The formula for calculating the iteration termination evaluation factor δ is as follows: In the above formula, ω0 is the set proportional coefficient of 0.7, and ω1 is the set proportional coefficient of 0.

3. If the evaluation factor δ does not fall within the interval [0,1] after the iteration ends, three points are randomly selected in the initial point cloud, and this step is repeated for the next iteration. If the evaluation factor δ falls within the interval [0,1] after the iteration ends, the iteration stops. After the iteration ends, the optimal model parameters are the extracted spatial plane equation. 6) Based on the type of weld on the target workpiece, determine the number of planes F of that type of weld. Following step 5), determine the first spatial plane equation. Separate the interior points on the determined first spatial plane equation from the three-dimensional point cloud data. Repeat step 5) on the remaining three-dimensional point cloud data to determine the same number of spatial plane equations as the number of planes F. 7) Based on the obtained F spatial plane equations, filter out the point cloud set within all spatial plane equations from the original laser profilometer scan data, and reconstruct the weld model; 8) From the reconstructed 3D weld model, determine the plane normal vectors of F spatial plane equations. Based on the input weld angle α, calculate the angle β between the plane normal vectors of the F spatial plane equations according to the direction of travel. Take the two normal vectors when Δt=|α-β| is the minimum, and determine the spatial plane equations corresponding to these normal vectors, which are the two weld side surfaces where the weld is located. 9) Obtain the equation of the intersecting line in space by extending the two planes in space. From the starting point of the scan to the ending point of the scan, obtain the line segment on the line equation, which is the weld position.

Images