Weld seam extraction method and apparatus, electronic device, and storage medium

Abstract

The present application discloses a weld seam extraction method and apparatus, an electronic device, and a storage medium. The method comprises: determining point cloud data of an area of interest under different illumination and different collection orientations (S110), wherein the area of interest is a welding area comprising a weld seam area; determining a weld seam point cloud data training set on the basis of the point cloud data of the area of interest (S120); training a deep learning network on the basis of the weld seam point cloud data training set to generate a weld seam extraction model (S130); and extracting weld seam point cloud data of a target welding area on the basis of the weld seam extraction model (S140).

Description

Weld extraction methods, devices, electronic equipment and storage media

[0001] This application claims priority to Chinese Patent Application No. 202410488928.2, filed with the Chinese Patent Office on April 23, 2024, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of weld quality inspection technology, such as a weld extraction method, apparatus, electronic device, and storage medium. Background Technology

[0003] Steel reinforcement is widely used in national infrastructure construction such as roads, buildings, and bridges. The primary method of connecting steel reinforcement is welding, and the quality of the welds determines the performance of the steel frame and, consequently, the safety and stability of the entire construction project. The dimensions of the steel welds are a crucial standard for judging weld quality. Currently, the main visual inspection method involves manual measurement using a measuring ruler. This method requires welding to be completed before inspection, resulting in low efficiency, heavy workload, and inaccurate results.

[0004] Summary of the Invention

[0005] This application provides a weld seam extraction method, apparatus, electronic device, and storage medium, which reduces the impact of environmental noise and lighting changes on the weld seam extraction method based on point cloud features, and increases the applicability of the method.

[0006] According to one aspect of this application, a method for extracting weld seams is provided, the method comprising:

[0007] Determine point cloud data of regions of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area;

[0008] Determine the weld seam point cloud data training set based on the point cloud data of the region of interest;

[0009] The deep learning network is trained based on the weld point cloud data training set to generate a weld extraction model;

[0010] The weld point cloud data of the target welding area is extracted based on the weld extraction model.

[0011] According to another aspect of this application, a weld extraction device is provided, the device comprising:

[0012] The data determination module is configured to determine point cloud data of the region of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area;

[0013] The training set determination module is configured to determine the weld point cloud data training set based on the point cloud data of the region of interest.

[0014] The model generation module is configured to train a deep learning network based on the weld point cloud data training set to generate a weld extraction model.

[0015] The weld extraction module is configured to extract weld point cloud data of the target welding area based on the weld extraction model.

[0016] According to another aspect of this application, an electronic device is provided, the electronic device comprising:

[0017] At least one processor; and

[0018] A memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the weld extraction method described in any embodiment of this application.

[0020] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the weld extraction method described in any embodiment of this application. Attached Figure Description

[0021] Figure 1 is a flowchart of a weld seam extraction method according to an embodiment of this application;

[0022] Figure 2 is a schematic diagram of manually annotated weld point cloud data according to an embodiment of this application;

[0023] Figure 3 is a schematic diagram of weld point cloud data extracted from a weld extraction model according to an embodiment of this application;

[0024] Figure 4 is a flowchart of a weld extraction method according to another embodiment of this application;

[0025] Figure 5 is a schematic diagram of a weld seam extraction device according to an embodiment of this application;

[0026] Figure 6 is a schematic diagram of the structure of an electronic device that implements the weld seam extraction method provided in an embodiment of this application. Detailed Implementation

[0027] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] Figure 1 is a flowchart of a weld seam extraction method according to an embodiment of this application. This embodiment is applicable to the extraction of weld seams in a welding area. The weld seam extraction method can be executed by a weld seam extraction device, which can be implemented in hardware and / or software and can be configured in an electronic device. As shown in Figure 1, the method includes:

[0029] S110. Determine the point cloud data of the region of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area.

[0030] In this embodiment, the Region of Interest (ROI) specifically refers to the welded area, including the weld seam area, which is crucial for weld quality inspection. Since the point cloud data acquired during acquisition includes point cloud data of the welded area under different lighting conditions and acquisition poses, it is also necessary to determine the point cloud data of the Region of Interest within the welded area.

[0031] For example, collecting point cloud data of the welded area under different lighting conditions and different acquisition poses includes: using a 3D laser scanner to collect point cloud data of the welded area under different lighting conditions; adjusting the pose relationship between the 3D laser scanner and the welded area to collect point cloud data of the welded area under different pose relationships.

[0032] Understandably, in order to collect point cloud data of the welded area under different lighting conditions and different acquisition poses, it is necessary to control the lighting conditions of the welded area and the pose relationship between the 3D laser scanner and the welded area when collecting point cloud data of the welded area. The former can be achieved by using an adjustable light source, and the latter can be achieved by adjusting the installation posture of the 3D laser scanner multiple times.

[0033] For example, determining the point cloud data of the region of interest under different lighting and different acquisition poses includes: denoising and downsampling the point cloud data of the welded area under different lighting and different acquisition poses; establishing a spatial rectangular coordinate system, and extracting the processed point cloud data based on the minimum and maximum coordinates of the weld area in the spatial rectangular coordinate system to obtain the point cloud data of the region of interest.

[0034] After acquiring point cloud data of the welded area under different lighting conditions and acquisition poses, the first step is to denoise and downsample the acquired point cloud data. This improves the quality of the point cloud data while reducing its quantity, thus increasing the speed and efficiency of subsequent processing. Secondly, the center of the welded area or another fixed point is selected as the origin of a spatial rectangular coordinate system. Based on the geometric characteristics of the welded area, the directions of the x, y, and z coordinate axes are determined, establishing a spatial rectangular coordinate system. Finally, the minimum and maximum values ​​of the x, y, and z axes of the weld area in the spatial rectangular coordinate system are determined. The denoised and downsampled point cloud data is then truncated to obtain the point cloud data of the region of interest. It is important to note that when truncating the point cloud data, the truncated area must completely cover the weld area, while avoiding the introduction of too much irrelevant point cloud data.

[0035] S120. Determine the weld seam point cloud data training set based on the point cloud data of the region of interest.

[0036] In this embodiment of the application, after determining the point cloud data of the region of interest under different lighting and different acquisition poses, a training set suitable for deep learning algorithms can be constructed through the feature points related to the weld in the point cloud data of the region of interest, namely the weld point cloud data training set.

[0037] For example, determining a weld point cloud data training set based on the point cloud data of the region of interest includes: determining the category features of the point cloud data of the region of interest; wherein the category features include weld, base material, and noise; and labeling the category features of the point cloud data of the region of interest to generate a weld point cloud data training set.

[0038] The weld seam is the primary target for identification and extraction, typically possessing a specific shape, size, and spatial distribution. The base material, the material surrounding the weld seam, usually appears as background and has different characteristics from the weld seam. Noise refers to irrelevant points introduced due to scanning equipment or environmental factors.

[0039] In this embodiment, the category features of point cloud data in the region of interest can be determined manually, and these category features are then input. Afterward, the category features of the point cloud data in the region of interest can be labeled. This process can be automated by computer or manually, and this embodiment does not limit this approach. For example, Figure 2 shows a schematic diagram of manually labeled weld point cloud data. Through the above steps, a high-quality weld point cloud data training set can be constructed, providing data support for subsequent tasks such as weld identification and extraction using deep learning algorithms.

[0040] S130. Train the deep learning network based on the weld point cloud data training set to generate a weld extraction model.

[0041] S140. Extract weld point cloud data of the target welding area based on the weld extraction model.

[0042] In this embodiment, since the processing involves point cloud data, specific deep learning networks designed for point cloud data processing, such as PointNet, PointNet++, and DGCNN, can be selected. These networks can directly process unordered point cloud data and extract useful features. The weld seam point cloud data training set is then input into the selected deep learning network. The network parameters are optimized using backpropagation and gradient descent algorithms, enabling the model to gradually learn the ability to extract weld seam point cloud data from point cloud data of regions of interest under different lighting conditions and acquisition poses. This generates a weld seam extraction model, which is then applied to the task of extracting weld seam point cloud data from the target welding area. For example, Figure 3 shows a schematic diagram of weld seam point cloud data extracted by a weld seam extraction model.

[0043] The technical solution of this application embodiment determines point cloud data of regions of interest (ROIs) under different lighting conditions and different acquisition poses; wherein the ROI is the welding area including the weld seam; a weld seam point cloud data training set is determined based on the ROI point cloud data; a deep learning network is trained using the weld seam point cloud data training set to generate a weld seam extraction model, and the weld seam point cloud data of the target welding area is extracted based on the weld seam extraction model. The technical solution of this application embodiment, by collecting weld seam point cloud data under different lighting conditions and different acquisition poses, creates a weld seam point cloud data training set containing a large amount of weld seam point cloud data to train the weld seam extraction model, reducing the impact of environmental noise and lighting changes on the weld seam extraction method based on point cloud features, and increasing the applicability of the method.

[0044] Figure 4 is a flowchart of a weld extraction method according to another embodiment of this application. This embodiment is a refinement based on the above embodiment; solutions not described in detail in this embodiment are described in the above embodiment. As shown in Figure 4, the method includes:

[0045] S210. Determine the point cloud data of the region of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area.

[0046] S220. Determine the weld seam point cloud data training set based on the point cloud data of the region of interest.

[0047] S230. Train the deep learning network based on the weld point cloud data training set to generate a weld extraction model.

[0048] In this embodiment of the application, since the weld point cloud data training set contains a large amount of point cloud data, in order to reduce the computational burden and improve the training efficiency, it is necessary to downsample or reduce the dimensionality of the point cloud data in the weld point cloud data training set during the training of the deep learning network based on the weld point cloud data training set.

[0049] For example, training a deep learning network to generate a weld extraction model based on the weld point cloud data training set includes: dividing the point cloud data in the weld point cloud data training set into cubes of uniform size; calculating the mean coordinates of the point cloud data within the cubes, and determining the category feature with the highest frequency among the category features corresponding to the point cloud data within the cubes, using the mean coordinates and the category feature as the features of the cubes; and treating the cubes as new point cloud data for the deep learning network to learn and generate a weld extraction model.

[0050] In this embodiment, the deep learning network further includes a downsampling module and a local feature aggregation and enhancement module. The downsampling module is configured to divide the point cloud data in the weld seam point cloud training set into cubes of uniform size. The local feature aggregation and enhancement module is configured to calculate the mean coordinates of the point cloud data within each cube and determine the most frequently occurring category feature among the category features corresponding to the point cloud data within each cube. The mean coordinates and the category feature are then used as the features of the cube. Through this process, the point cloud data in the weld seam point cloud training set can be transformed into cubes containing category features of the point cloud data. Compared to the original point cloud data, the new cubes not only have a significantly reduced number of points but also contain sufficient information for subsequent deep learning tasks. Therefore, the cubes can be regarded as new point cloud data for the deep learning network to learn and generate a weld seam extraction model, thereby reducing the computational burden and improving training efficiency.

[0051] S240. Adjust the 3D laser scanner so that the weld area of ​​the target welding area is always within the measurement range of the 3D laser scanner.

[0052] In this embodiment, before acquiring point cloud data of the target welding area, the pose relationship between the 3D laser scanner and the target welding area can be coarsely calibrated. For example, by adjusting the mounting posture of the 3D laser scanner, the weld area of ​​the target welding area before and after welding is always within the measurement range of the 3D laser scanner. To ensure the accuracy of point cloud data acquisition, a test acquisition of the target welding area can be performed before formal acquisition to verify whether the adjustment of the 3D laser scanner is effective and whether further adjustments are needed.

[0053] S250. By estimating the minimum and maximum coordinates of the weld area in the target welding area in the spatial rectangular coordinate system, the target region of interest in the target welding area is determined.

[0054] In this embodiment, after coarsely calibrating the pose relationship between the 3D laser scanner and the target welding area, the target region of interest in the target welding area can be determined, for example, by estimating the minimum and maximum coordinates of the weld area in the spatial rectangular coordinate system. For instance, the center of the target welding area or another fixed point is selected as the origin of the spatial rectangular coordinate system, and the directions of the x, y, and z coordinate axes are determined based on the geometric characteristics of the target welding area to establish the spatial rectangular coordinate system. Then, the minimum and maximum coordinates of the weld area along the x, y, and z axes in the spatial rectangular coordinate system are estimated to determine the target region of interest in the target welding area.

[0055] S260. Extract weld point cloud data of the target welding area based on the weld extraction model.

[0056] For example, extracting weld point cloud data of the target welding area based on the weld model includes: acquiring point cloud data of the target region of interest using a 3D laser scanner, preprocessing the point cloud data of the target region of interest, and inputting the preprocessed data into the deep learning model to extract the weld point cloud data of the target welding area.

[0057] Understandably, in step S250, the target region of interest (ROI) within the target welding area has already been determined. To reduce computational load and further improve the efficiency of extracting weld point cloud data from the target welding area, point cloud data of the ROI can be directly acquired using a 3D laser scanner. After preprocessing the point cloud data of the ROI, it is input into a deep learning model for extracting the weld point cloud data of the target welding area. The preprocessing includes denoising and downsampling the point cloud data of the ROI.

[0058] S270. Using principal component analysis of point cloud data, calculate the length, width, and depth of the weld point cloud data.

[0059] Principal component analysis (PCA) is a statistical method that uses orthogonal transformations to convert linearly correlated variables in the original feature space into new linearly independent variables, known as principal components. In this embodiment, PCA can be used to determine the main directions of the weld point cloud data, thereby determining the length, width, and depth of the weld.

[0060] The technical solution of this application embodiment determines point cloud data of the region of interest (ROI) under different lighting and acquisition poses; wherein, the ROI is the welding area including the weld seam area; a weld seam point cloud data training set is determined based on the ROI point cloud data; a deep learning network is trained on the weld seam point cloud data training set to generate a weld seam extraction model; the 3D laser scanner is adjusted so that the weld seam area of ​​the target welding area is always within the measurement range of the 3D laser scanner; the target ROI in the target welding area is determined by estimating the minimum and maximum coordinates of the weld seam area in the spatial rectangular coordinate system; the weld seam point cloud data of the target welding area is extracted based on the weld seam extraction model; and the length, width, and depth of the weld seam point cloud data are calculated using principal component analysis of the point cloud data. The technical solution of this application embodiment, by collecting weld seam point cloud data under different lighting and acquisition poses and creating a weld seam point cloud data training set containing a large amount of weld seam point cloud data to train the weld seam extraction model, reduces the impact of environmental noise and lighting changes on the weld seam extraction method based on point cloud features, and increases the applicability of the method. Meanwhile, by pre-determining the target region of interest, the computational load is reduced and the efficiency of weld seam point cloud data extraction is improved.

[0061] Figure 5 is a schematic diagram of a weld seam extraction device provided in an embodiment of this application. As shown in Figure 5, the device includes:

[0062] The data determination module 310 is configured to determine point cloud data of the region of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is a welding area including the weld seam area;

[0063] The training set determination module 320 is configured to determine a weld point cloud data training set based on the point cloud data of the region of interest.

[0064] The model generation module 330 is configured to train a deep learning network based on the weld point cloud data training set to generate a weld extraction model.

[0065] The weld extraction module 340 is configured to extract weld point cloud data of the target welding area based on the weld extraction model.

[0066] For example, data determination module 310 includes:

[0067] The denoising and downsampling unit is configured to perform denoising and downsampling processing on the point cloud data of the welded area under different lighting conditions and different acquisition poses.

[0068] The data determination unit is set to establish a spatial rectangular coordinate system. Based on the minimum and maximum coordinates of the weld area in the spatial rectangular coordinate system, the processed point cloud data is truncated to obtain the point cloud data of the region of interest.

[0069] For example, training set determination module 320 includes:

[0070] The category feature determination unit is configured to determine the category features of the point cloud data of the region of interest; wherein, the category features include weld seam, base material, and noise.

[0071] The training set determination unit is configured to label the category features of the point cloud data of the region of interest and generate a training set of weld point cloud data.

[0072] For example, model generation module 330 includes:

[0073] The point cloud data segmentation unit is configured to segment the point cloud data in the weld point cloud data training set into cubes of the same size.

[0074] The cube feature determination unit is configured to calculate the mean coordinates of the point cloud data within the cube, and determine the category feature with the highest frequency among the category features corresponding to the point cloud data within the cube, and use the mean coordinates and the category feature as the features of the cube.

[0075] The model generation unit is configured to treat the cube as new point cloud data and use it for the deep learning network to learn and generate a weld extraction model.

[0076] For example, the device further includes:

[0077] The coarse calibration module is set to adjust the 3D laser scanner so that the weld area of ​​the target welding area is always within the measurement range of the 3D laser scanner.

[0078] The target region of interest determination module is configured to determine the target region of interest in the target welding area by estimating the minimum and maximum coordinates of the weld area in the spatial rectangular coordinate system.

[0079] For example, weld extraction module 340 includes:

[0080] The preprocessing unit is configured to collect point cloud data of the target region of interest using a 3D laser scanner and preprocess the point cloud data of the target region of interest.

[0081] The weld seam point cloud data extraction unit is configured to input the preprocessed data into the deep learning model to extract the weld seam point cloud data of the target welding area.

[0082] For example, the device further includes:

[0083] The weld seam point cloud data calculation module is configured to use principal component analysis of point cloud data to calculate the length, width, and depth of the weld seam point cloud data.

[0084] The weld extraction device provided in this application embodiment can execute the weld extraction method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects of the method execution.

[0085] Figure 6 illustrates a schematic diagram of the structure of an electronic device 10 that can be used to implement embodiments of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0086] As shown in Figure 6, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0087] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0088] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs several of the methods and processes described above, such as weld seam extraction methods.

[0089] In some embodiments, the weld seam extraction method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the weld seam extraction method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the weld seam extraction method by any other suitable means (e.g., by means of firmware).

[0090] The various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard parts (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0091] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0092] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), optical fibers, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. A computer-readable storage medium can be a non-transitory computer-readable storage medium.

[0093] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a cathode ray tube (CRT) or liquid crystal display (LCD) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0094] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0095] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system. It avoids the management difficulties and weak business scalability associated with physical hosts and Virtual Private Servers (VPS) in related technologies.

[0096] It should be understood that the various processes shown above can be used to rearrange, add, or delete steps. For example, the multiple steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

Claims

1. A method for extracting weld seams, comprising: Determine point cloud data of regions of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area; Determine the weld seam point cloud data training set based on the point cloud data of the region of interest; The deep learning network is trained based on the weld point cloud data training set to generate a weld extraction model; The weld point cloud data of the target welding area is extracted based on the weld extraction model.

2. The method according to claim 1, wherein, The point cloud data used to determine the region of interest under different lighting conditions and different acquisition poses includes: Denoising and downsampling processing were performed on the point cloud data of the welded area under different lighting conditions and different acquisition poses. A spatial rectangular coordinate system is established. The processed point cloud data is then truncated based on the minimum and maximum coordinate values ​​of the weld area in the spatial rectangular coordinate system to obtain the point cloud data of the region of interest.

3. The method according to claim 1, wherein, The step of determining the weld point cloud data training set based on the point cloud data of the region of interest includes: Determine the category features of the point cloud data of the region of interest; wherein, the category features include weld seam, base material, and noise. The category features of the point cloud data of the region of interest are labeled to generate the weld seam point cloud data training set.

4. The method according to claim 1, wherein, The step of training a deep learning network based on the weld point cloud data training set to generate a weld extraction model includes: The point cloud data in the weld seam point cloud data training set is divided into cubes of uniform size. Calculate the mean coordinates of the point cloud data within the cube, and determine the category feature with the highest frequency among the category features corresponding to the point cloud data within the cube. Use the mean coordinates and the category feature with the highest frequency as the features of the cube. The cube is treated as new point cloud data and used by the deep learning network to learn in order to generate the weld extraction model.

5. The method according to claim 1, further comprising, before extracting the weld point cloud data of the target welding area based on the weld extraction model: Adjust the 3D laser scanner so that the weld area of ​​the target welding area is always within the measurement range of the 3D laser scanner; The target region of interest within the target welding area is determined by estimating the minimum and maximum coordinates of the weld area in the spatial rectangular coordinate system.

6. The method according to claim 5, wherein, The extraction of weld point cloud data of the target welding area based on the weld model includes: The point cloud data of the target region of interest is acquired by the three-dimensional laser scanner, and the point cloud data of the target region of interest is preprocessed. The preprocessed point cloud data is input into the deep learning model to extract the weld point cloud data of the target welding area.

7. The method according to claim 1, after extracting the weld point cloud data of the target welding area based on the weld model, further comprising: The length, width, and depth of the weld point cloud data are calculated using principal component analysis of point cloud data.

8. A weld seam extraction device, comprising: The data determination module is configured to determine point cloud data of the region of interest under different lighting conditions and different acquisition poses; wherein, the region of interest is the welding area including the weld seam area; The training set determination module is configured to determine the weld point cloud data training set based on the point cloud data of the region of interest. The model generation module is configured to train a deep learning network based on the weld point cloud data training set to generate a weld extraction model. The weld extraction module is configured to extract weld point cloud data of the target welding area based on the weld extraction model.

9. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the weld extraction method according to any one of claims 1-7.

10. A computer-readable storage medium storing computer instructions that, when executed by a processor, implement the weld extraction method according to any one of claims 1-7.

Images