Workpiece welding seam stable recognition and positioning method achieved through multi-assembly point cloud registration

Abstract

The invention discloses a workpiece welding seam stable recognition and positioning method achieved through multi-assembly point cloud registration. The method comprises the following specific steps that S1, welding assembly point cloud is obtained and preprocessed; s2, processing the point cloud by using a KD-Tree, and complementing a missing part; s3, the welding assemblies are welded into a whole, and multi-view scanning is carried out; s4, performing registration on the multi-view point cloud through coarse registration and fine registration; s5, point cloud splicing is carried out, and a complete welding assembly point cloud model is obtained; and S6, extracting a welding seam, removing a pseudo welding seam, and finally identifying an actual welding seam. According to the welding seam stable identification method based on low-overlap point cloud registration, through the multi-view scanning and low-overlap point cloud registration technology, high registration precision can be kept under the low-overlap condition, the welding seam identification stability and accuracy are remarkably improved, the welding seam position can be automatically identified, pseudo welding seams can be removed, and the welding seam identification efficiency is improved. Errors caused by traditional manual detection are avoided, and the production efficiency and the welding quality are improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of computer vision, point cloud processing and welding technology, and in particular to a method for stably identifying and locating a workpiece weld by multi-component point cloud registration. Background Art

[0002] In the field of industrial welding manufacturing, accurate detection and identification of welds constitute the core steps of teaching-free technology. With the continuous advancement of science and technology, the technical solution of automatic defect detection of welds using advanced equipment such as depth perception cameras, complementary metal oxide semiconductor (CMOS) image sensors and lidar has become a key topic in the current research and development field. At present, teaching-free welding technology faces several significant challenges in practice, especially when dealing with multi-workpiece weld identification tasks. Existing algorithms often encounter problems such as misjudgment of welds and errors in weld position identification when dealing with weld identification. These problems directly affect the accuracy and efficiency of welding.

[0003] In order to improve the accuracy of weld recognition, the industry has already developed a solution that introduces a three-dimensional model of the workpiece as an auxiliary means to optimize the recognition and positioning of the weld through model information. This method can theoretically significantly improve the accuracy of recognition, but its prerequisite is that the three-dimensional model of the workpiece must be obtained in advance. This step not only increases the amount of preparatory work for teaching-free welding, but also increases the complexity of the entire process, making the popularization and application of the technology face certain obstacles.

[0004] Introducing workpiece models to assist in weld identification provides a feasible idea for improving weld identification accuracy. However, how to simplify the operating process and lower the technical threshold while ensuring accuracy is still a key issue that needs to be urgently solved in the future development of teaching-free welding technology. To address these issues, the present invention proposes a workpiece weld stable identification and positioning method achieved by multi-component point cloud registration. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the shortcomings of the existing technology, the present invention provides a method for stable identification and positioning of workpiece welds achieved by multi-component point cloud registration, which solves the following problems: the operating procedures of the existing technology are complicated and the threshold is high, and the existing point cloud registration methods mostly rely on high-overlap or strong alignment point cloud data. In the case of low overlap, the point cloud registration accuracy and stability are poor, which affects the accuracy of weld identification.

[0007] (II) Technical solution

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] The technical solution of the present invention is: a method for stable identification and positioning of workpiece welds by multi-component point cloud registration, comprising the following steps:

[0010] S1, obtain the point cloud of the welding component and perform preprocessing;

[0011] S2, use KD-Tree to process the point cloud and complete the missing parts;

[0012] S3, welding the welding components into a whole and performing multi-view scanning;

[0013] S4, align the multi-view point clouds through coarse and fine registration;

[0014] S5, performing point cloud stitching to obtain a complete point cloud model of the welding component;

[0015] S6, extract the welds and remove the pseudo welds, and finally identify the actual welds.

[0016] As a preferred technical solution of the present invention, in step S1, point cloud data is obtained, and the welding assembly is scanned using a line structure laser camera and other equipment to obtain its three-dimensional point cloud data. The point cloud data is subjected to welding platform point cloud ablation, point cloud denoising, and down sampling to remove noise and reduce redundant data.

[0017] As a preferred technical solution of the present invention, in step S2, point cloud preprocessing and missing parts completion are performed, and the point cloud of the welding assembly is processed using KD-Tree. KD-Tree is used to quickly find the nearest neighbor points, perform range search, and perform point cloud segmentation, etc. The point cloud data is classified into a cloud set of points to be welded, the missing areas in the point cloud of each workpiece are identified, and the coordinates and attributes of the points around these missing areas are calculated. The missing parts are completed using a linear interpolation algorithm to obtain a complete point cloud of the welding assembly.

[0018] As a preferred technical solution of the present invention, in step S3, the point cloud is scanned from multiple perspectives to spot weld the welding components into an integral workpiece to be welded. Then, the welded structure is scanned from multiple perspectives using equipment such as an array laser camera to obtain multi-angle point cloud data. The obtained multi-angle point cloud data is pre-processed by denoising and removing redundancy to ensure the subsequent point cloud alignment and stitching effects.

[0019] As a preferred technical solution of the present invention, in step S4, point cloud registration is performed, wherein coarse registration: first, a coarse registration stage is performed, and when the initial conditions are unknown, a point cloud registration matrix is ​​quickly estimated by a method based on local feature description, a global search strategy or a statistical probability method; fine registration: on the basis of the coarse registration, fine registration is performed by an iterative nearest point algorithm to optimize the transformation matrix obtained by the coarse registration, and finally an accurate point cloud registration result is obtained.

[0020] As a preferred technical solution of the present invention, in step S5, point cloud splicing is performed by matching the registered point cloud data, calculating the correspondence between the point clouds, and finding their feature similarity. Finally, the point cloud data from multiple perspectives are unified into the same coordinate system and spliced ​​into a complete point cloud data model.

[0021] As a preferred technical solution of the present invention, in step S6, weld identification is performed to extract the welds to be welded. The three-dimensional coordinate information of the acquired point cloud data is used to analyze the geometric shapes of different point cloud patches, calculate the normal vector of each patch, and determine whether adjacent patches intersect. If so, the intersection line is obtained. Based on the workpiece point cloud data initially acquired, the intersection lines of the intersection patches of each workpiece are calculated, and the pseudo welds inherent in the workpiece itself are eliminated. The remaining intersection lines are the actual welds.

[0022] (III) Beneficial effects

[0023] The present invention provides a method for stable identification and positioning of workpiece welds by multi-component point cloud registration. It has the following beneficial effects:

[0024] (1) There is no need to import the model of the workpiece to be welded, which greatly reduces the tedious preparatory work for teaching-free welding and improves work efficiency.

[0025] (2) Avoiding misjudgment of welds: Through high-precision source point cloud and target point cloud low-overlap registration technology, it is possible to effectively avoid misidentifying the inherent structure of the welded component itself (such as the bend of a bent part) as a weld, thereby ensuring the accuracy of weld identification.

[0026] (3) Accurate weld positioning: By using the intersection relationship between point cloud patches, the exact position of the actual weld is calculated, which reduces the error of weld identification and improves the accuracy and quality of welding. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] See also Figure 1The embodiment of the present invention provides a technical solution: a method for stable identification and positioning of workpiece welds based on multi-component point cloud registration. The method for stable identification of multi-workpiece welds based on low-overlap point cloud registration described in the present invention includes: point cloud acquisition and preprocessing: using a laser camera to perform point cloud scanning to obtain point cloud data on the surface of the welding component. By preprocessing the scanned point cloud data, such as sampling, denoising, and ablating the welding platform, the subsequent processing process is simplified.

[0030] Point cloud completion and processing: When there are missing areas in the point cloud data, KD-Tree is used to quickly find points around the missing parts, and linear interpolation is used to fill in the missing points to ensure the integrity of the point cloud data.

[0031] Multi-view point cloud scanning and registration: Scan the welded structure from multiple viewpoints to ensure that the point cloud data covers the entire welding area. Use the coarse registration algorithm to preliminarily align the point clouds from different viewpoints, and then use the ICP algorithm for fine registration to ensure the consistency of the point cloud in three-dimensional space.

[0032] Point cloud stitching and weld extraction: stitch the point cloud data from each perspective into a complete point cloud model, and extract the suspected weld position by calculating the normal vector and intersection analysis of the facet. Remove false welds by comparing with the geometric model of the workpiece itself.

[0033] The specific method steps are as follows: use a line structure laser camera and other equipment to scan the welding assembly, obtain the point cloud of the welding assembly, perform welding platform point cloud ablation, point cloud noise reduction, down sampling and other operations on the point cloud of the welding assembly, so as to improve the efficiency and accuracy of subsequent processing;

[0034] KD-Tree is used to process the point cloud of welding components. KD-Tree can be used for tasks such as quickly finding the nearest neighbor points, performing range searches, and segmenting point clouds. The acquired point clouds are classified into cloud sets of points to be welded, and the coordinates and attributes of the points around the missing areas of each workpiece point cloud are calculated. The missing parts of the weldment point cloud are completed through a linear interpolation algorithm.

[0035] The welding components are spot welded into an integral workpiece to be welded, namely a welding structure. The welding structure is scanned from multiple perspectives using an array laser camera and other equipment to obtain multi-perspective point clouds. The obtained point clouds are then processed (such as point cloud denoising) for subsequent operations.

[0036] Perform multi-view point cloud registration. In the point cloud registration stage, point cloud data acquired from different viewpoints or positions are aligned to the same coordinate system to ensure that they correspond correctly in space. This is divided into two stages, namely coarse registration and fine registration of point clouds. Coarse registration is to quickly estimate an approximate point cloud registration matrix under the condition of unknown initial conditions. Common coarse registration algorithms include methods based on local feature description, global search strategies, and statistical probability methods; fine registration uses the initial transformation matrix obtained by coarse registration to obtain a more accurate solution through calculations such as the iterative closest point algorithm (ICP algorithm).

[0037] The purpose of stitching multi-view point clouds is to merge the previously registered point cloud data into a complete point cloud model. The key to stitching is to find the correspondence between the initial point cloud and the target point cloud, match them according to the correspondence, calculate their feature similarity, and finally unify them into a coordinate system to synthesize a complete point cloud data.

[0038] Extract the welds to be welded, analyze the three-dimensional coordinate information of the acquired point cloud data, identify different point cloud patches, calculate the normal vector of each patch, and then determine whether they intersect and find the intersection line, which is suspected to be the weld. Then, based on the point cloud data of each component obtained at the beginning, calculate the intersection line of its intersecting patches, remove the identical intersection lines (pseudo welds inherent in the component itself), and the remaining intersection lines are the welds actually required to be welded.

[0039] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0040] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for stable identification and positioning of workpiece welds by multi-component point cloud registration, characterized in that: The specific steps include the following: S1, obtain the point cloud of the welding component and perform preprocessing; S2, use KD-Tree to process the point cloud and complete the missing parts; S3, welding the welding components into a whole and performing multi-view scanning; S4, align the multi-view point clouds through coarse and fine registration; S5, performing point cloud stitching to obtain a complete point cloud model of the welding component; S6, extract the welds and remove the pseudo welds, and finally identify the actual welds.

2. The method for stable identification and positioning of workpiece welds by multi-component point cloud registration according to claim 1, characterized in that: In step S1, point cloud data is obtained, and the welding component is scanned using a line structure laser camera or other equipment to obtain its three-dimensional point cloud data. The point cloud data is subjected to welding platform point cloud ablation, point cloud denoising, and down sampling to remove noise and reduce redundant data.

3. The method for stable identification and positioning of workpiece welds by multi-component point cloud registration according to claim 1, characterized in that: In step S2, the point cloud is preprocessed and the missing parts are completed. The point cloud of the welding component is processed using KD-Tree. KD-Tree is used to quickly find the nearest neighbor points, perform range search, and segment the point cloud. The point cloud data is classified into a cloud set of points to be welded, the missing areas in the point cloud of each workpiece are identified, and the coordinates and attributes of the points around these missing areas are calculated. The missing parts are completed using a linear interpolation algorithm to obtain a complete point cloud of the welding component.

4. The method for stable identification and positioning of workpiece welds by multi-component point cloud registration according to claim 1, characterized in that: In step S3, the point cloud is scanned from multiple perspectives to spot weld the welding components into an integral workpiece to be welded. Then, the welded structure is scanned from multiple perspectives using a planar array laser camera and other equipment to obtain point cloud data from multiple perspectives. The obtained multi-perspective point cloud data is preprocessed by denoising and removing redundancy to ensure the subsequent point cloud registration and stitching effects.

5. The method for stable identification and positioning of workpiece welds by multi-component point cloud registration according to claim 1, characterized in that: In step S5, point cloud splicing is performed by matching the registered point cloud data, calculating the correspondence between point clouds, and finding their feature similarity. Finally, the point cloud data from multiple perspectives are unified into the same coordinate system and spliced ​​into a complete point cloud data model.

6. The method for stable identification and positioning of workpiece welds by multi-component point cloud registration according to claim 1, characterized in that: In step S6, the weld is identified and the weld to be welded is extracted. The geometric shapes of different point cloud patches are analyzed using the three-dimensional coordinate information of the acquired point cloud data, the normal vector of each patch is calculated, and it is determined whether adjacent patches intersect. If they do intersect, the intersection line is obtained. Based on the workpiece point cloud data initially acquired, the intersection lines of the intersection patches of each workpiece are calculated, and the pseudo welds inherent in the workpiece itself are eliminated. The remaining intersection lines are the actual welds.

Images