Method and apparatus for spray coating protective tooling of magnetic material substrate

Abstract

This application provides a method and apparatus for processing protective coating fixtures for magnetic material substrates, relating to the field of coating fixture technology. The method includes: acquiring grayscale images, depth maps, and point cloud images of a target area of ​​the magnetic material substrate; performing filtering, feature extraction, and feature fusion on the grayscale images, depth maps, and point cloud images respectively to obtain a three-dimensional fused feature point cloud; performing region connectivity processing to extract the three-dimensional spatial distribution features of the target area; reconstructing the surface based on the three-dimensional spatial distribution features to obtain a reconstructed surface model of the target area; meshing the reconstructed surface model to generate a meshed damage model; unfolding the meshed damage model in two dimensions to obtain a two-dimensional flattened model; converting the two-dimensional flattened model into a laser cutting file; and performing laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the protective coating fixture for the target area. This application significantly improves the efficiency and accuracy of coating repair on magnetic material substrates.

Description

Technical Field

[0001] This application relates to the field of spraying tooling technology, specifically to a method and apparatus for processing spraying protective tooling on a magnetic material substrate. Background Technology

[0002] In the field of spray coating protection processing for magnetic material substrates, the current mainstream protection solutions in the industry still rely heavily on traditional manual operation mode, that is, manually cutting tape and applying it around the damaged area of ​​the magnetic substrate, exposing only the damaged part to be sprayed for repair in order to achieve local protection.

[0003] Traditional manual methods present numerous challenges when applying protective coatings to complex curved aircraft skins. Firstly, the damage patterns on aircraft surfaces are often irregular, making it difficult to precisely match the damaged edges when manually cutting tape. Secondly, the adhesion pressure and flatness between the tape and the substrate surface are difficult to control evenly during manual application, easily leading to defects such as bubbles and wrinkles. Furthermore, manually applying tape requires individual positioning, cutting, bonding, and subsequent removal, resulting in low efficiency and failing to meet the demands for rapid skin repair. The repetitive and meticulous nature of these tasks over extended periods can cause operator fatigue, further reducing coating accuracy and efficiency. Therefore, there is an urgent need to address the low accuracy and efficiency issues when applying protective coatings to complex curved magnetic material substrates. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the prior art, this application provides a method and apparatus for processing protective coating on magnetic material substrates, which effectively solves the problem of low spraying accuracy and efficiency when processing protective coating on magnetic material substrates with complex curved surfaces.

[0005] In a first aspect, this application provides a method for processing a protective coating on a magnetic material substrate using a spray coating tooling, the method comprising:

[0006] Acquire grayscale, depth, and point cloud images of the target region on the magnetic material substrate;

[0007] Filtering and feature extraction are performed on the grayscale image, the depth image, and the point cloud image respectively to obtain grayscale edge information, depth edge information, and point cloud edge set;

[0008] The grayscale edge information, the depth edge information, and the point cloud edge set are fused to obtain a three-dimensional fused feature point cloud;

[0009] Based on the three-dimensional fused feature point cloud, regional connectivity processing is performed to extract the three-dimensional spatial distribution features of the target region;

[0010] Surface reconstruction is performed based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target region.

[0011] The reconstructed surface model is meshed to generate a meshed damage model;

[0012] The meshed damage model is unfolded in two dimensions to obtain a two-dimensional flattened model;

[0013] The two-dimensional flattened model is converted into a laser cutting file, and the magnetic material workpiece is laser-cut according to the laser cutting file to obtain the spraying protective fixture for the target area.

[0014] In an optional implementation, filtering and feature extraction are performed on the grayscale image, the depth image, and the point cloud image respectively to obtain grayscale edge information, depth edge information, and a set of point cloud edges, including:

[0015] A Gabor filter is used to enhance the edges of a grayscale image to obtain a filtered grayscale image. The local maximum value of the Gabor filter response is calculated based on the filtered grayscale image to obtain the grayscale edge information.

[0016] Gaussian filtering is applied to the depth map to obtain a filtered depth map, and the image gradient and non-maximum suppression are calculated based on the filtered depth map to determine the edges, thereby obtaining the depth edge information;

[0017] The point cloud image is denoised using a random sampling consensus algorithm to obtain a filtered point cloud image, and principal component analysis is performed on the filtered point cloud image to obtain the point cloud edge set.

[0018] In an optional implementation, the grayscale edge information, the depth edge information, and the point cloud edge set are fused to obtain a three-dimensional fused feature point cloud, including:

[0019] The grayscale edge information and the depth edge information are mapped to three-dimensional space using camera intrinsic parameters and camera extrinsic parameters, respectively, to obtain three-dimensional grayscale edge point cloud and three-dimensional depth edge point cloud;

[0020] The three-dimensional grayscale edge point cloud, the three-dimensional depth edge point cloud, and the point cloud edge set are weighted and fused to obtain a three-dimensional fused feature point cloud.

[0021] In an optional implementation, the grayscale edge information and the depth edge information are mapped to three-dimensional space using camera intrinsic parameters and camera extrinsic parameters respectively, to obtain three-dimensional grayscale edge point clouds and three-dimensional depth edge point clouds, including:

[0022] Extract edge pixels from the grayscale edge information and the depth edge information respectively to obtain grayscale pixel coordinates and depth pixel coordinates;

[0023] The grayscale pixel coordinates and the depth pixel coordinates are transformed to the normalized plane in the camera coordinate system using the camera intrinsic parameters, respectively, to obtain the grayscale pixel camera coordinates and the depth pixel camera coordinates.

[0024] Based on the depth value corresponding to the pixel camera coordinates in the camera coordinate system, the grayscale pixel camera coordinates and the depth pixel camera coordinates are converted into pixel coordinates in the three-dimensional camera coordinate system to obtain the grayscale pixel three-dimensional camera coordinates and the depth pixel three-dimensional camera coordinates.

[0025] The grayscale pixel 3D camera coordinates and the depth pixel 3D camera coordinates are transformed to the world coordinate system using the camera extrinsic parameters to obtain the 3D grayscale edge point cloud and the 3D depth edge point cloud.

[0026] In an optional implementation, region connectivity processing is performed based on the 3D fused feature point cloud to extract the 3D spatial distribution features of the target region, including:

[0027] The three-dimensional fused feature point cloud is spatially meshed to obtain a three-dimensional voxel structure;

[0028] Extract the local maxima of the gradient in the three-dimensional voxel structure to obtain the preliminary three-dimensional edge;

[0029] The gradient magnitude of the preliminary three-dimensional edge is filtered using a dual threshold method to obtain the first target edge and the second target edge;

[0030] The first target edge and the second target edge are connected by a region growing algorithm to obtain the three-dimensional edge of the target;

[0031] Based on connected component analysis, the independent edge segments of the target's three-dimensional edge are identified to obtain the three-dimensional spatial distribution characteristics.

[0032] In an optional implementation, surface reconstruction is performed based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target region, including:

[0033] The gradient magnitude of each voxel in the three-dimensional spatial distribution features is obtained, and the gradient magnitude is linearly normalized to obtain the normalized magnitude.

[0034] The normalized amplitude is then subjected to a nonlinear transformation to obtain a nonlinear amplitude.

[0035] Isosurfaces are extracted based on the nonlinear amplitude, and initial triangular patches are generated based on the isosurfaces;

[0036] The initial triangular facet is optimized using a quality error metric to obtain the target triangular facet.

[0037] The target triangular facets are connected and converted into a target format to obtain the reconstructed surface model.

[0038] In an optional implementation, the reconstructed surface model is meshed to generate a meshed damage model, including:

[0039] The vertex set of the reconstructed surface model is subjected to Delaunay triangulation to obtain a preliminary triangular mesh;

[0040] Calculate the quadratic error matrix for each vertex of the initial triangular mesh, and construct an edge priority queue sorted by shrinkage cost based on the quadratic error matrix;

[0041] The edge with the minimum shrinkage cost is selected from the edge priority queue for shrinkage operation, the mesh topology is updated, and the target triangular mesh is obtained.

[0042] The target triangular mesh is filtered to obtain the meshed damage model.

[0043] In an optional implementation, the meshed damage model is unfolded in two dimensions to obtain a two-dimensional flattened model, including:

[0044] Construct a graphical Laplacian matrix based on the meshed damage model;

[0045] Calculate the eigenvalues ​​and eigenvectors of the graph Laplacian matrix;

[0046] Select a target number of non-trivial feature values ​​and non-trivial feature vectors based on the feature values ​​and feature vectors, respectively.

[0047] A two-dimensional embedding is constructed based on the non-trivial eigenvalues ​​and the non-trivial eigenvectors to obtain a two-dimensional point set;

[0048] The two-dimensional point set is optimized by conformal mapping and damage constraint terms are added to obtain the two-dimensional flattened model.

[0049] In an optional implementation, converting the two-dimensional flattened model into a laser cutting file includes:

[0050] Based on the two-dimensional flattened model, the boundary is extracted, the outer contour is separated, and the outer contour separation file is obtained.

[0051] The external contour separation file is converted to a different format to obtain the laser cutting file.

[0052] Secondly, this application provides a spray coating protective tooling processing apparatus for a magnetic material substrate, the apparatus comprising:

[0053] The image acquisition module is used to acquire grayscale images, depth maps, and point cloud images of the target area on the magnetic material substrate.

[0054] The image processing module is used to perform filtering and feature extraction on the grayscale image, the depth image and the point cloud image respectively to obtain grayscale edge information, depth edge information and point cloud edge set;

[0055] The feature fusion module is used to fuse the grayscale edge information, the depth edge information, and the point cloud edge set to obtain a three-dimensional fused feature point cloud;

[0056] The feature extraction module is used to perform region connectivity processing based on the three-dimensional fused feature point cloud to extract the three-dimensional spatial distribution features of the target region;

[0057] The model building module is used to reconstruct the surface based on the three-dimensional spatial distribution features to obtain a reconstructed surface model of the target area.

[0058] The model processing module is used to perform meshing processing on the reconstructed surface model to generate a meshed damage model;

[0059] The model flattening module is used to perform two-dimensional unfolding of the meshed damage model to obtain a two-dimensional flattened model.

[0060] The laser cutting module is used to convert the two-dimensional flattened model into a laser cutting file, and to perform laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the spraying protective fixture for the target area.

[0061] The method and apparatus for processing protective coating fixtures for magnetic material substrates provided in this application achieve high-precision identification and modeling of damaged areas of magnetic material substrates by fusing grayscale images, depth maps, and point cloud images and performing three-dimensional spatial alignment and weighted fusion. This enables precise positioning of damaged areas on complex curved magnetic material substrates before coating. An improved manifold learning-based spectral flattening algorithm is employed to unfold the complex curved surface damage model into a two-dimensional plane with low distortion, ensuring the geometric fidelity of the fixture cutting. Combining laser cutting and magnetic adsorption characteristics, rapid customization and precise fitting of the protective coating fixture are achieved, significantly improving the efficiency and consistency of coating repair. Simultaneously, a closed-loop processing mechanism is implemented from damage area perception to protective coating fixture generation, greatly reducing human intervention and operational errors. This method is suitable for the efficient and precise coating protection needs of complex curved magnetic material substrates such as aircraft skin, and has promising engineering application prospects. Attached Figure Description

[0062] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 This is a first schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0064] Figure 2 This is a second schematic diagram of the process flow for the spraying protective tooling for a magnetic material substrate provided in the embodiments of this application;

[0065] Figure 3 This is a third schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0066] Figure 4 This is the fourth schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0067] Figure 5 This is the fifth schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0068] Figure 6 This is the sixth schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0069] Figure 7 This is the seventh schematic diagram of the process flow of the spraying protective tooling for magnetic material substrates provided in the embodiments of this application;

[0070] Figure 8 This is a schematic diagram of the structure of the spraying and protective tooling processing device for magnetic material substrates provided in the embodiments of this application;

[0071] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0072] Key component symbols: 200 - Spray coating and protective tooling processing device for magnetic material substrate; 210 - Image acquisition module; 220 - Image processing module; 230 - Feature fusion module; 240 - Feature extraction module; 250 - Model building module; 260 - Model processing module; 270 - Model flattening module; 280 - Laser cutting module; 300 - Electronic equipment; 310 - Processor; 320 - Communication interface; 330 - Memory; 340 - Communication bus. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be further described clearly and completely below with reference to the accompanying drawings of the embodiments. It should be noted that the described embodiments are merely some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0074] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0075] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0076] During long-term service, the magnetic substrate material skin of aircraft often suffers localized damage, even detachment, due to bird strikes, sand erosion, corrosion, or fatigue cracks. Current mainstream protection solutions in the industry still heavily rely on traditional manual methods, involving manually cutting and applying tape around the damaged area of ​​the magnetic substrate, exposing only the damaged area to be sprayed for repair to achieve localized protection.

[0077] However, traditional manual operations present numerous problems when processing protective coatings on complex curved aircraft skins. Damage to aircraft surfaces is often irregular, making it difficult to precisely match the edges of damage when manually cutting tape. Furthermore, the pressure and flatness of the tape against the substrate are difficult to control evenly during manual application, easily leading to defects such as bubbles and wrinkles. The precision of manual tape application heavily relies on operator experience, requiring individual positioning, cutting, application, and subsequent removal, resulting in low efficiency and difficulty meeting the needs for rapid skin repair. Moreover, prolonged repetitive and meticulous operations can easily lead to operator fatigue, further reducing the precision and efficiency of the protective coating. Therefore, there is an urgent need to address the problem of low precision and efficiency in protective coating processes on complex curved surfaces and magnetic material substrates.

[0078] This application provides a method for processing a protective coating tooling for a magnetic material substrate, which effectively solves the problem of low spraying accuracy and efficiency when processing protective coatings for magnetic material substrates such as complex curved surfaces. Figure 1This is a first schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown below. Figure 1 As shown, the method includes the following steps:

[0079] S100: Obtain grayscale images, depth maps, and point cloud maps of the target region on the magnetic material substrate.

[0080] In this embodiment, the magnetic material substrate can be the skin of an aircraft, and the target area is the damaged area of ​​the skin. A 3D industrial camera can be used to acquire grayscale images, depth maps, and point cloud images of the damaged area. The grayscale image provides surface texture and visual feature information of the damaged area; the depth map reflects the height distribution of points on the surface of the damaged area and is used to capture geometrical abrupt changes such as pits and peeling caused by the damage; and the point cloud image is a set of three-dimensional coordinates generated by structured light scanning, which fully describes the spatial morphology of the damaged area and its surrounding surfaces.

[0081] In the embodiments of this application, grayscale images, depth maps, and point cloud images respectively carry visual and geometric information of different modalities. The three work together to provide multi-dimensional data support for achieving high-precision three-dimensional modeling and edge extraction of irregular damaged areas on magnetic material substrates.

[0082] S200. Filter and extract features from the grayscale image, depth image, and point cloud image respectively to obtain grayscale edge information, depth edge information, and point cloud edge set.

[0083] In this embodiment, image preprocessing is performed on the grayscale image, depth map, and point cloud image to extract relevant feature information. Figure 2 This is a second schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown below. Figure 2 As shown, image preprocessing specifically includes the following steps:

[0084] S210. Use a Gabor filter to enhance the edges of the grayscale image to obtain a filtered grayscale image, and calculate the local maximum value of the Gabor filter response based on the filtered grayscale image to obtain grayscale edge information.

[0085] As an optional implementation of this application, a Gabor filter can be applied to the grayscale image to enhance texture edges in specific directions, highlighting linear damage features such as cracks and scratches, thus obtaining a filtered grayscale image. The Gabor filter combines spatial localization and frequency selectivity, suppressing irrelevant background interference and improving the edge signal-to-noise ratio, providing high-quality input for subsequent accurate extraction of grayscale edge information.

[0086] Then, by calculating the local maximum value of the Gabor filter response on the filtered grayscale image, the pixel with the most drastic texture change can be accurately located, thereby extracting direction-sensitive edge features, effectively preserving significant edges consistent with the damage contour, suppressing noise response, generating high-confidence grayscale edge information, and providing reliable two-dimensional clues for multimodal edge fusion.

[0087] S220. Perform Gaussian filtering on the depth map to obtain a filtered depth map, and calculate the image gradient and non-maximum suppression based on the filtered depth map to determine the edges and obtain depth edge information.

[0088] As an optional implementation of this application, a Gauss filter can be used to smooth noise in the depth map, suppressing random errors in industrial camera image acquisition while retaining key geometric boundary information to obtain a filtered depth map. This method improves the continuity and stability of depth data, avoids false edges caused by noise, and ensures that the extracted depth edge information accurately reflects the actual deformation contour.

[0089] Furthermore, Canny edge detection is performed on the filtered depth map. First, the image gradient magnitude and direction are calculated to identify regions with drastic depth changes. Then, non-maximum suppression is used to preserve local maxima of the gradient, accurately locating edge pixels and eliminating blurred boundaries. This process effectively extracts the contours of abrupt depth changes caused by pits and cracks, generating clear and closed depth edge information, providing high-precision geometric basis for damage boundary localization in three-dimensional space.

[0090] S230. The point cloud image is denoised using a random sampling consensus algorithm to obtain a filtered point cloud image. Principal component analysis is then performed on the filtered point cloud image to obtain a set of point cloud edges.

[0091] As an optional implementation method of this application, the Random Sampling Consensus (RANSAC) algorithm can be used to denoise the point cloud image. This algorithm automatically fits local geometric models such as planes or curved surfaces, treating points that conform to the model as interior points and discarding points that deviate significantly as noise, thus obtaining a filtered point cloud image. This algorithm is robust, effectively removing outliers and abnormal points caused by industrial camera errors, preserving the true surface structure, and significantly improving the quality of the point cloud.

[0092] Furthermore, principal component analysis (PCA) is performed on the filtered point cloud image. By analyzing the covariance matrix of local neighborhood points, the main direction of the point cloud distribution is identified. In the edge regions, the point cloud exhibits obvious directional differences, and eigenvalue decomposition can be used to determine whether it is located at the boundary. Based on this, points with significant changes in principal components are extracted and clustered to form a continuous set of point cloud edges, effectively preserving the geometric contour features in three-dimensional space and supporting subsequent multimodal feature fusion.

[0093] S300: The grayscale edge information, depth edge information, and point cloud edge set are fused to obtain a three-dimensional fused feature point cloud.

[0094] In this embodiment of the application, the two-dimensional grayscale edge information and depth edge information are mapped to three-dimensional space based on the intrinsic and extrinsic parameters of the industrial camera, and multimodal feature fusion is performed with the point cloud edge set to obtain a three-dimensional fused feature point cloud. Figure 3 This is a third schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown in the figure. Figure 3 As shown, the multimodal feature fusion specifically includes the following steps:

[0095] S310. Using camera intrinsic parameters and camera extrinsic parameters, grayscale edge information and depth edge information are mapped to three-dimensional space respectively to obtain three-dimensional grayscale edge point cloud and three-dimensional depth edge point cloud.

[0096] In this embodiment, each edge pixel in the grayscale edge information is first extracted to obtain grayscale pixel coordinates. Then, the camera intrinsic parameter matrix is ​​obtained based on the camera intrinsic parameters. The inverse of the camera intrinsic parameter matrix is ​​used to transform the grayscale pixel coordinates to a normalized plane in the camera coordinate system, resulting in grayscale pixel camera coordinates in the camera coordinate system. Next, combined with the depth value corresponding to the grayscale pixel camera coordinates in the camera coordinate system, the grayscale pixel camera coordinates in the camera coordinate system are converted to pixel coordinates in the 3D camera coordinate system, obtaining grayscale pixel 3D camera coordinates. Finally, the grayscale pixel 3D camera coordinates in the 3D camera coordinate system are transformed to the world coordinate system using camera extrinsic parameters, forming a 3D grayscale edge point cloud with grayscale gradient intensity.

[0097] Similarly, back-projection calculations are performed on the depth edge information: First, each edge pixel in the depth edge information is extracted to obtain the depth pixel coordinates. Then, the inverse of the camera intrinsic matrix is ​​used to transform the depth pixel coordinates to a normalized plane in the camera coordinate system, obtaining the depth pixel camera coordinates in the camera coordinate system. Next, combined with the depth value corresponding to the depth pixel camera coordinates in the camera coordinate system, the depth pixel camera coordinates in the camera coordinate system are transformed into pixel coordinates in the 3D camera coordinate system, obtaining the depth pixel 3D camera coordinates. Finally, the depth pixel 3D camera coordinates in the 3D camera coordinate system are transformed to the world coordinate system using camera extrinsic parameters, forming a 3D depth edge point cloud with depth gradient intensity.

[0098] S320. Weighted fusion of the 3D grayscale edge point cloud, the 3D depth edge point cloud, and the point cloud edge set is performed to obtain the 3D fused feature point cloud.

[0099] In this embodiment, based on the principle of spatial proximity, for each point in the point cloud edge set, its spatially nearest point can be found in both the 3D grayscale edge point cloud and the 3D depth edge point cloud, respectively, to obtain the nearest neighbor point. Then, the point cloud edge feature intensity of each point in the point cloud edge set, the grayscale gradient intensity of the nearest point in the 3D grayscale edge point cloud, and the depth gradient intensity of the nearest point in the 3D depth edge point cloud are obtained.

[0100] Then, a weighted fusion strategy is adopted based on nearest neighbor points. The point cloud edge feature intensity, gray-level gradient intensity, and depth gradient intensity are weighted and averaged according to preset weight ratios to obtain the fused feature intensity. Among them, the feature intensity corresponding to the point cloud edge set usually has the highest weight, and the gradient intensity corresponding to the 3D gray-level edge point cloud and the 3D depth edge point cloud has a relatively balanced weight. Optionally, the preset weights of the point cloud edge feature intensity, gray-level gradient intensity, and depth gradient intensity can be set to 0.2, 0.3, and 0.5, respectively.

[0101] Finally, after selecting the nearest neighbor and performing weighted fusion calculations on each point in the point cloud edge set, a 3D fused feature point cloud containing accurate 3D coordinates and fused feature intensity is output. The accurate 3D coordinates can be obtained from the point cloud edge set.

[0102] This application embodiment acquires different edge point cloud datasets through camera parameters, establishes the correspondence between different datasets based on nearest neighbor matching, and generates a comprehensive intensity value through a weighted average with preset weights. Finally, it outputs a high-quality 3D fusion feature point cloud that integrates the advantages of vision, depth perception, and 3D geometry. This 3D fusion feature point cloud integrates multimodal edge information and has good robustness and accuracy.

[0103] S400. Perform region connectivity processing based on the 3D fusion feature point cloud to extract the 3D spatial distribution features of the target region.

[0104] In this embodiment of the application, the gradient of the three-dimensional fusion feature is calculated by a gradient-based three-dimensional edge detection algorithm. The local maxima of the gradient are the three-dimensional edges. Regional connectivity analysis is performed on the three-dimensional edges to obtain the three-dimensional spatial distribution features of the skin damage area. Figure 4 This is the fourth schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown below. Figure 4 As shown, the region connectivity processing specifically includes the following steps:

[0105] S410. Spatial meshing is performed on the 3D fused feature point cloud to obtain a 3D voxel structure.

[0106] In this embodiment of the application, the space is uniformly divided into countless tiny cubic grids through spatial meshing, and the discrete and irregularly distributed three-dimensional fusion feature point cloud is converted into a regular three-dimensional voxel grid structure. Each voxel contains the fusion feature intensity value at its corresponding position, thereby constructing a continuous three-dimensional scalar field and obtaining a three-dimensional voxel structure.

[0107] S420. Extract the local maxima of the gradient in the three-dimensional voxel structure to obtain the preliminary three-dimensional edge.

[0108] As an optional implementation of this application, a three-dimensional Sobel operator can be used to calculate the gradient of the three-dimensional scalar field. Corresponding three-dimensional convolution kernels are applied in the x, y, and z directions, and local partial derivatives are calculated voxel by voxel to obtain the gradient components on the three coordinate axes. Then, the gradient magnitude at each voxel is obtained by calculating the magnitude of the gradient vector, i.e., the square root of the sum of the squares of the partial derivatives in each direction. This gradient magnitude reflects the drastic change in the intensity of local features and is a key basis for judging three-dimensional edges; the larger the gradient magnitude, the more likely it is located at the edge of the skin damage region.

[0109] Furthermore, for each voxel, the gradient magnitude is interpolated and compared along the gradient direction. If the gradient magnitude of the current voxel is greater than its two nearest neighbors along the gradient direction, it is determined to be a local maximum gradient magnitude. The 3D voxel structure of the local maximum gradient magnitude is preserved to obtain the preliminary 3D edge. Non-maximum suppression is used to eliminate spurious responses, thereby forming a preliminary 3D edge with more accurate localization and clearer structure, laying the foundation for subsequent dual-threshold detection and edge connection.

[0110] S430. The gradient magnitude of the preliminary three-dimensional edge is screened using the double threshold method to obtain the first target edge and the second target edge.

[0111] As an optional implementation of this application, a dual-threshold method can be used to filter the gradient magnitude of the preliminary 3D edges. This involves setting two gradient magnitude thresholds: a first threshold and a second threshold, where the first threshold is greater than the second threshold. The preliminary 3D edges after non-maximum suppression are classified according to their gradient magnitude: points with gradient magnitudes higher than the first threshold are identified as first target edges, i.e., strong edges, and are considered highly likely to be true edges. Points with gradient magnitudes between the first and second thresholds are identified as first target edges, i.e., weak edges, which may be true edges but require further verification. Points with gradient magnitudes lower than the second threshold are directly discarded as noise or irrelevant details.

[0112] S440. Connect the edges of the first target and the second target using a region growing algorithm to obtain the three-dimensional edge of the target.

[0113] In this embodiment, a first target edge is used as a seed point to extend and connect to a neighboring second target edge. Starting from each first target edge point, the second target edge points in its neighborhood are traversed. If a connection path exists, these second target edge points are incorporated into the final edge to obtain the target's three-dimensional edge. This process effectively fills in edge breaks caused by noise or gradient decay, connecting originally discontinuous edge segments into a complete contour.

[0114] S450. Based on connected component analysis, identify independent edge segments of the target's three-dimensional edge to obtain three-dimensional spatial distribution characteristics.

[0115] In this embodiment, connected component analysis is used to cluster voxels in the target 3D edge according to spatial connectivity, grouping adjacent edge voxels into the same group. Based on the 26-neighborhood connectivity criterion, when the current voxel has an edge response and is connected to any of its 26 neighboring voxels, it is considered as the same connected region, thus effectively identifying independent edge segments that are separated from each other. Connected component analysis can distinguish multiple fracture regions or branch structures in complex damaged edges, avoid false connections, ensure the spatial integrity of each independent edge segment, and finally obtain the 3D spatial distribution characteristics of each independent edge segment. The 3D spatial distribution characteristics include multiple voxels, and the gradient magnitude of each voxel constitutes the voxel data source.

[0116] This application embodiment suppresses noise and connects fracture edges through processing methods such as nonmaximum suppression, dual threshold analysis, region growth, and connected component analysis, ensuring the continuity and reliability of the three-dimensional edges. The resulting three-dimensional spatial distribution features accurately express the spatial location, contour direction, and geometric shape of the skin damage area, providing a high-quality data foundation for subsequent surface reconstruction and tooling generation, and significantly improving the accuracy and efficiency of spray protection operations.

[0117] S500: Reconstruct the surface based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target area.

[0118] In this embodiment of the application, a reconstructed surface model is obtained by reconstructing the surface based on the three-dimensional spatial distribution features. Figure 5 This is the fifth schematic diagram of the process flow for the spraying protective tooling for a magnetic material substrate provided in this application embodiment, as shown below. Figure 5 As shown, surface reconstruction specifically includes the following steps:

[0119] S510. Obtain the gradient magnitude of each voxel in the three-dimensional spatial distribution features, and linearly normalize the gradient magnitude to obtain the normalized magnitude.

[0120] In this embodiment, the gradient magnitude of each voxel in the three-dimensional spatial distribution features is obtained, and the gradient magnitude reflects the significance of its location at the edge. By linear normalization, the gradient magnitudes of all voxels are mapped to the standard interval [0,1] to obtain normalized magnitudes, thereby eliminating the influence of dimensions and ensuring the consistency of data distribution.

[0121] S520. Perform a nonlinear transformation on the normalized amplitude to obtain a nonlinear amplitude.

[0122] As an optional implementation of this application, a nonlinear transformation of the normalized amplitude can be performed using a sigmoid function to enhance edge contrast and suppress noise. The sigmoid function has the characteristics of smooth transition and saturation at both ends: for strong edge regions close to 1, the output value is further compressed to approach 1; for weak edges and background regions close to 0, the output value is pulled down to approach 0; and intermediate values ​​are moderately stretched. This nonlinear transformation enhances the difference between significant edges and non-edge regions, making edges sharper and clearer.

[0123] S530. Extract the isosurface based on the nonlinear amplitude, and generate an initial triangular patch based on the isosurface.

[0124] In this embodiment, the entire three-dimensional voxel structure is traversed, and the nonlinear amplitudes of all voxels are collected. These scalar values ​​are divided into several intervals to generate a histogram of gradient amplitude distribution. The first peak of this histogram corresponds to the background and non-edge regions, the second peak corresponds to the edge structure, and the valley between the two peaks serves as the nonlinear amplitude threshold for distinguishing between edges and non-edges.

[0125] Based on a nonlinear amplitude threshold, the range of voxels with the same nonlinear amplitude is defined as an isosurface. Voxels are used as vertices to construct voxel cells, each consisting of eight vertices and a corresponding nonlinear amplitude. For each voxel, the relationship between the nonlinear amplitudes of its eight vertices and the isosurface determines whether the isosurface passes through the voxel. For voxels that pass through, isosurface points are interpolated on the edges based on the vertex states and connected to form triangular patches, generating the corresponding initial triangular patches.

[0126] S540. The initial triangular facet is optimized using a quality error measurement method to obtain the target triangular facet.

[0127] In this embodiment, a quality error metric method can be used to optimize the initial triangular facets. Specifically, a quadratic error matrix is ​​defined for each vertex of the initial triangular facet to quantify the geometric changes in its neighborhood. During the simplification process, the edges that minimize the geometric error are preferentially selected for shrinkage, i.e., replacing the two endpoints with a new vertex and updating the mesh topology to obtain the target triangular facet. This process automatically preserves the details of areas with high curvature such as edges and corners, while significantly reducing redundant triangles in flat areas, thus reducing the number of facets and improving surface quality.

[0128] S550: Connect the target triangular facets and convert them into a target format to obtain the reconstructed surface model.

[0129] In this embodiment, the target triangular facets are connected to obtain an optimized surface, and the optimized surface is exported as an STL format to obtain a reconstructed surface model.

[0130] The embodiments of this application efficiently reconstruct discrete three-dimensional spatial distribution features into continuous and closed triangular mesh surfaces, accurately restore the complex geometry of the skin damage area, and achieve high-fidelity and low-redundancy surface modeling through isosurface extraction and quality error measurement optimization.

[0131] S600: Mesh the reconstructed surface model to generate a meshed damage model.

[0132] In this embodiment, Delaunay triangulation is used to mesh the reconstructed surface model to generate a meshed damage model. Figure 6 This is the sixth schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown in the figure. Figure 6 As shown, the meshing process specifically includes the following steps:

[0133] S610. Perform Delaunay triangulation on the vertex set of the reconstructed surface model to obtain a preliminary triangular mesh.

[0134] In this embodiment, Delaunay triangulation is performed on the vertex set of the reconstructed surface model. Discrete vertices in three-dimensional space are constructed into a set of non-overlapping and topologically sound triangular faces using geometric algorithms, forming a preliminary triangular mesh. This process follows the principle that the circumcircle of any triangle does not contain other vertices, thereby maximizing the minimum angle, avoiding elongated triangles, and improving mesh quality.

[0135] S620. Calculate the quadratic error matrix of each vertex of the initial triangular mesh, and construct an edge priority queue sorted by shrinkage cost based on the quadratic error matrix.

[0136] In this embodiment, a quadratic error matrix is ​​calculated for each vertex of the initial triangular mesh. The quadratic error matrix consists of the normal vector of the plane where the vertex is located and the distance from the point to the plane. It accumulates to represent the local shape information of the surrounding geometric surface and is used to quantify the geometric importance of the vertex's location.

[0137] When shrinking an edge, its two endpoints are merged into a new vertex. The shape deviation caused by the edge shrinkage is evaluated using this quadratic error matrix. The shrinkage cost of each edge is calculated by determining the quadratic error matrix generated after merging the two endpoints of the edge into a new vertex, and obtaining the optimal vertex coordinates that minimize the quadratic error, thus preserving the original geometric features to the greatest extent. All edges are sorted according to their corresponding minimum quadratic error values, constructing an edge priority queue arranged in ascending order of shrinkage cost. This ensures that edges with the least impact on the overall geometry are shrunk first during the simplification process, achieving high-fidelity mesh simplification.

[0138] S630. Select the edge with the minimum shrinkage cost according to the edge priority queue, perform the shrinkage operation, update the mesh topology, and obtain the target triangular mesh.

[0139] In this embodiment, iterative simplification is performed based on an edge priority queue. Each time, the edge with the smallest shrinkage cost is selected from the edge priority queue for shrinkage operation. The two vertices of the edge are merged into a new vertex, the mesh topology is updated, and the resulting degenerate triangles are removed to obtain the target triangular mesh.

[0140] S640. Filter the target triangular mesh to obtain a meshed damage model.

[0141] As an optional implementation of this application, the target triangular mesh can be filtered to smooth geometric noise, eliminate minute jagged edges, and improve the continuity of the triangular facets, while preserving the key contour features of the skin damage area. For example, the filtering process can employ algorithms such as Laplacian filtering or bilateral filtering to suppress high-frequency errors while maintaining edge sharpness. The filtered meshed damage model is then generated in STL format, exhibiting higher geometric accuracy and topological quality.

[0142] S700: Perform two-dimensional unfolding on the meshed damage model to obtain a two-dimensional flattened model.

[0143] In this embodiment of the application, the meshed damage model is subjected to spectral flattening based on improved manifold learning to obtain a two-dimensional flattened model. Figure 7 This is the seventh schematic diagram of the process flow for the spraying protective tooling of the magnetic material substrate provided in the embodiments of this application, as shown in the diagram. Figure 7 As shown, the two-dimensional unfolding specifically includes the following steps:

[0144] S710. Construct the graph Laplacian matrix based on the gridded damage model.

[0145] In this embodiment of the application, information from a meshed damage model in STL format, including vertex sets, is read. and target triangular mesh set , n This indicates the number of target triangular meshes in the meshed damage model. v i Indicates the first i The vertices of the target triangular mesh, f i Indicates the first i A target triangular mesh.

[0146] In the meshed damage model, each target triangular mesh is treated as a graph, with each vertex as a node and the triangle edges as edges. The graph Laplacian matrix is ​​then calculated. To display the connectivity and structure of a graph, the graph Laplace matrix is ​​used. The expression is as follows:

[0147]

[0148] In the above formula, i and j Indicates the index of the target triangular mesh. v i Indicates the first i The vertices of the target triangular mesh, v j Indicates the first j The vertices of the target triangular mesh, E Represents the set of adjacent vertices, otherwise means if v i and v j Not adjacent.

[0149] S720. Calculate the eigenvalues ​​and eigenvectors of the graph Laplacian matrix.

[0150] In this embodiment, the eigenvalues ​​and eigenvectors of each graph Laplacian matrix are solved using the following formula:

[0151]

[0152] In the above formula, Represents eigenvalues. l This represents the corresponding feature vector. D The degree matrix is ​​a diagonal matrix. D The diagonal elements are ,in The adjacency weight is defined as the sum of the damage similarity factors and the edge weights. The sum of the cotangent angles of two adjacent triangles, and their adjacency weight. The calculation formula is as follows:

[0153]

[0154] In the above formula, and Representing an edge Two opposite angles, d i and d j They represent the first i The vertices of the target triangular mesh and the first j Damage values ​​of the vertices of the target triangular mesh. This indicates the control of damage sensitivity.

[0155] S730. Select the target number of non-trivial eigenvalues ​​and non-trivial eigenvectors based on the eigenvalues ​​and eigenvectors respectively.

[0156] In this embodiment of the application, the first eigenvalue and eigenvector are selected from all eigenvalues ​​and eigenvectors. Eigenvalues ​​of nontrivial eigenvectors and the corresponding feature vectors , serving as non-trivial eigenvalues ​​and non-trivial eigenvectors. Optional, k The value can be set according to the actual situation.

[0157] S740. Construct a two-dimensional embedding based on non-trivial eigenvalues ​​and non-trivial eigenvectors to obtain a two-dimensional point set.

[0158] In this embodiment, each three-dimensional vertex is mapped to its respective two-dimensional point using non-trivial eigenvalues ​​and non-trivial eigenvectors, resulting in a spectral flattened two-dimensional point set of the meshed damage model. The mapping method is as follows:

[0159]

[0160] In the above formula, u i Indicates the first i The two-dimensional points corresponding to the vertices of the target triangular mesh. l 1i and l 2i Indicates the first i The first and second non-trivial eigenvectors corresponding to the target triangular mesh.

[0161] S750. Perform conformal mapping optimization on the two-dimensional point set and add damage constraint terms to obtain a two-dimensional flattened model.

[0162] In this embodiment of the application, conformal mapping optimization is performed on the two-dimensional point set to minimize angular distortion, resulting in an optimized two-dimensional point set. The conformal mapping optimization is as follows:

[0163]

[0164] In the above formula, Indicates the adjacency weight. u i and u j They represent the first i The and the first j The two-dimensional points corresponding to the vertices of the target triangular mesh.

[0165] Furthermore, damage constraints are added to maintain the spatial proximity of damage similarities, ultimately resulting in a two-dimensional flattened model. The added damage constraints are as follows:

[0166]

[0167] In the above formula, m and n Indicates the penalty coefficient. Indicates the minimum allowable distance. τ This represents the threshold for damage similarity.

[0168] In this embodiment of the application, the final output is a two-dimensional flattened model in OBJ format.

[0169] In this embodiment, an improved manifold learning-based spectral flattening algorithm is used for the meshed damage model. The topological structure is extracted through the graph Laplacian matrix, and the damage-sensitive weights and non-trivial feature mapping are combined to effectively suppress distortion and wrinkles in the large-area flattening process. Damage similarity constraints and conformal optimization are introduced to further maintain the spatial proximity and contour continuity of key areas, achieving low-distortion and tear-free two-dimensional unfolding. It can unfold complex surfaces into a two-dimensional plane with high fidelity while preserving local geometric features.

[0170] S800: Convert the two-dimensional flattened model into a laser cutting file, and perform laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the spraying protective fixture for the target area.

[0171] In this embodiment, the boundary extraction can be performed by opening the two-dimensional flattened model using CAD software to separate the outer contour, thereby accurately separating the boundary of the damaged area, forming a closed and continuous geometric path, and obtaining an outer contour separation file. This outer contour separation file ensures the integrity and adaptability of the tooling fitting edge.

[0172] Then, the format of the external contour separation file is converted to the DXF format, a common format for laser cutting machines. Finally, the magnetic material workpiece is laser-cut using the laser cutting file containing skin damage information, resulting in the cut magnetic material workpiece, which serves as the spraying protection fixture for the skin damage area. This spraying protection fixture is then attached to the skin damage area to facilitate subsequent spraying.

[0173] This application embodiment utilizes a laser cutting file containing damage information to perform laser cutting on magnetic material workpieces, which can produce a spraying protective fixture with high precision that matches the actual damage morphology, ensuring smooth cuts and accurate dimensions. This enables a rapid conversion from three-dimensional damage modeling to two-dimensional physical fixtures, significantly improving the positioning accuracy, fit, and work efficiency of spraying protection, and meeting the needs for in-situ and rapid repair.

[0174] The coating protection tooling processing method for magnetic material substrates provided in this application achieves high-precision identification and modeling of damaged areas of the magnetic material substrate by fusing grayscale images, depth maps, and point cloud images and performing three-dimensional spatial alignment and weighted fusion. This enables precise positioning of damaged areas on complex curved magnetic material substrates before coating. An improved manifold learning-based spectral flattening algorithm is employed to unfold the complex curved surface damage model into a two-dimensional plane with low distortion, ensuring the geometric fidelity of tooling cutting. Combining laser cutting and magnetic adsorption characteristics, rapid customization and precise fitting of the coating protection tooling are achieved, significantly improving the efficiency and consistency of coating repair.

[0175] Based on the same technical concept, embodiments of this application also provide a spraying protective tooling processing device for magnetic material substrates. Figure 8 This is a schematic diagram of the structure of the spraying and protective tooling processing device for the magnetic material substrate provided in the embodiments of this application, as shown below. Figure 8 As shown, the spray coating protective tooling processing device 200 for the magnetic material substrate includes:

[0176] Image acquisition module 210 is used to acquire grayscale image, depth image and point cloud image of target area of ​​magnetic material substrate.

[0177] The image processing module 220 is used to perform filtering and feature extraction on the grayscale image, depth image and point cloud image respectively to obtain grayscale edge information, depth edge information and point cloud edge set.

[0178] The feature fusion module 230 is used to fuse grayscale edge information, depth edge information and point cloud edge set to obtain a three-dimensional fused feature point cloud.

[0179] The feature extraction module 240 is used to perform regional connectivity processing based on the 3D fused feature point cloud to extract the 3D spatial distribution features of the target region.

[0180] The model building module 250 is used to reconstruct the surface based on the three-dimensional spatial distribution characteristics and obtain the reconstructed surface model of the target area.

[0181] The model processing module 260 is used to perform meshing processing on the reconstructed surface model to generate a meshed damage model.

[0182] The model flattening module 270 is used to perform two-dimensional unfolding of the meshed damage model to obtain a two-dimensional flattened model.

[0183] The laser cutting module 280 is used to convert a two-dimensional flattened model into a laser cutting file, and to perform laser cutting on a magnetic material workpiece according to the laser cutting file to obtain a spraying protective fixture for the target area.

[0184] The magnetic material substrate spraying protection tooling processing device provided in this application embodiment realizes a closed-loop process from damage area perception to spraying protection tooling generation, which greatly reduces human intervention and operation error. It is suitable for the efficient and precise spraying protection needs of complex curved magnetic material substrates such as aircraft skin, and has good engineering application prospects.

[0185] It is understood that the implementation method of the spraying protective tooling processing method of the magnetic material substrate in the above embodiment is also applicable to this embodiment and can achieve the same technical effect, so it will not be described again here.

[0186] Based on the same concept, this application also provides an electronic device. Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 9 As shown, the electronic device 300 may include a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute the steps of the spray coating protective tooling processing method for the magnetic material substrate as described in the above embodiments. For example, it includes:

[0187] S100: Obtain grayscale image, depth map and point cloud image of the target area of ​​the magnetic material substrate;

[0188] S200. Filter and extract features from the grayscale image, depth image and point cloud image respectively to obtain grayscale edge information, depth edge information and point cloud edge set;

[0189] S300: The grayscale edge information, depth edge information, and point cloud edge set are fused to obtain a three-dimensional fused feature point cloud;

[0190] S400. Perform region connectivity processing based on the 3D fusion feature point cloud to extract the 3D spatial distribution features of the target region.

[0191] S500: Reconstruct the surface based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target area;

[0192] S600: Mesh the reconstructed surface model to generate a meshed damage model;

[0193] S700: Perform two-dimensional unfolding on the meshed damage model to obtain a two-dimensional flattened model;

[0194] S800: Convert the two-dimensional flattened model into a laser cutting file, and perform laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the spraying protective fixture for the target area.

[0195] The processor 310 can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0196] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0197] The memory 330 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, the memory may include high-speed random access memory and non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0198] Based on the same concept, embodiments of this application also provide a computer-readable storage medium storing a computer program containing at least one piece of code executable by a master control device to control the master control device to implement the steps of the spraying protective tooling processing method for the magnetic material substrate as described in the above embodiments. For example, it includes:

[0199] S100: Obtain grayscale image, depth map and point cloud image of the target area of ​​the magnetic material substrate;

[0200] S200. Filter and extract features from the grayscale image, depth image and point cloud image respectively to obtain grayscale edge information, depth edge information and point cloud edge set;

[0201] S300: The grayscale edge information, depth edge information, and point cloud edge set are fused to obtain a three-dimensional fused feature point cloud;

[0202] S400. Perform region connectivity processing based on the 3D fusion feature point cloud to extract the 3D spatial distribution features of the target region.

[0203] S500: Reconstruct the surface based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target area;

[0204] S600: Mesh the reconstructed surface model to generate a meshed damage model;

[0205] S700: Perform two-dimensional unfolding on the meshed damage model to obtain a two-dimensional flattened model;

[0206] S800: Convert the two-dimensional flattened model into a laser cutting file, and perform laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the spraying protective fixture for the target area.

[0207] Based on the same technical concept, this application also provides a computer program, which, when executed by a main control device, is used to implement the above-described method embodiments.

[0208] The computer program may be stored, in whole or in part, on a computer-readable storage medium packaged with the processor, or in part or in whole on a memory not packaged with the processor.

[0209] Based on the same technical concept, this application also provides a processor for implementing the above-described method embodiments. The processor can be a chip.

[0210] In summary, the coating protection tooling processing method and apparatus for magnetic material substrates provided in this application achieves high-precision identification and modeling of damaged areas of magnetic material substrates by fusing grayscale images, depth maps, and point cloud images and performing three-dimensional spatial alignment and weighted fusion. This enables precise positioning of damaged areas on complex curved magnetic material substrates before coating. An improved manifold learning-based spectral flattening algorithm is employed to unfold the complex curved surface damage model into a two-dimensional plane with low distortion, ensuring the geometric fidelity of tooling cutting. Combining laser cutting and magnetic adsorption characteristics, rapid customization and precise fitting of the coating protection tooling are achieved, significantly improving the efficiency and consistency of coating repair. Simultaneously, a closed-loop processing mechanism is implemented from damage area perception to coating protection tooling generation, greatly reducing human intervention and operational errors. This method is suitable for the efficient and precise coating protection needs of complex curved magnetic material substrates such as aircraft skin, and has promising engineering application prospects.

[0211] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0212] The embodiments described above are merely examples of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these modifications and improvements all fall within the protection scope of this application.

[0213] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for processing a protective coating on a magnetic material substrate, characterized in that, The method includes: Acquire grayscale, depth, and point cloud images of the target region on the magnetic material substrate; Filtering and feature extraction are performed on the grayscale image, the depth image, and the point cloud image respectively to obtain grayscale edge information, depth edge information, and point cloud edge set; The grayscale edge information, the depth edge information, and the point cloud edge set are fused to obtain a three-dimensional fused feature point cloud; Based on the three-dimensional fused feature point cloud, region connectivity processing is performed to obtain the three-dimensional spatial distribution features of the target region; Surface reconstruction is performed based on the three-dimensional spatial distribution characteristics to obtain a reconstructed surface model of the target region. The reconstructed surface model is meshed to generate a meshed damage model; The meshed damage model is unfolded in two dimensions to obtain a two-dimensional flattened model; The two-dimensional flattened model is converted into a laser cutting file, and the magnetic material workpiece is laser-cut according to the laser cutting file to obtain the spraying protective fixture for the target area. The step of reconstructing the surface based on the three-dimensional spatial distribution features to obtain a reconstructed surface model of the target region includes: The gradient magnitude of each voxel in the three-dimensional spatial distribution features is obtained, and the gradient magnitude is linearly normalized to obtain the normalized magnitude. The normalized amplitude is then subjected to a nonlinear transformation to obtain a nonlinear amplitude. Isosurfaces are extracted based on the nonlinear amplitude, and initial triangular patches are generated based on the isosurfaces; The initial triangular facet is optimized using a quality error metric to obtain the target triangular facet. The target triangular facets are connected and converted into a target format to obtain the reconstructed surface model; The step of unfolding the meshed damage model into a two-dimensional flattened model includes: Construct a graphical Laplacian matrix based on the meshed damage model; Calculate the eigenvalues ​​and eigenvectors of the graph Laplacian matrix; Select a target number of non-trivial feature values ​​and non-trivial feature vectors based on the feature values ​​and feature vectors, respectively. A two-dimensional embedding is constructed based on the non-trivial eigenvalues ​​and the non-trivial eigenvectors to obtain a two-dimensional point set; The two-dimensional point set is optimized by conformal mapping and damage constraint terms are added to obtain the two-dimensional flattened model. The damage constraint terms are added as follows: In the above formula, u Represents a two-dimensional point set, u i Indicates the first i The two-dimensional points corresponding to the vertices of the target triangular mesh. u j Indicates the first j The two-dimensional points corresponding to the vertices of the target triangular mesh. m and n Indicates the penalty coefficient. Indicates the minimum allowable distance. d i and d j They represent the first i The vertices of the target triangular mesh and the first j Damage values ​​of the vertices of the target triangular mesh. τ This represents the threshold for damage similarity.

2. The method for processing a magnetic material substrate using spray coating protective tooling according to claim 1, characterized in that, Filtering and feature extraction are performed on the grayscale image, the depth image, and the point cloud image respectively to obtain grayscale edge information, depth edge information, and a set of point cloud edges, including: A Gabor filter is used to enhance the edges of a grayscale image to obtain a filtered grayscale image. The local maximum value of the Gabor filter response is calculated based on the filtered grayscale image to obtain the grayscale edge information. Gaussian filtering is applied to the depth map to obtain a filtered depth map, and the image gradient and non-maximum suppression are calculated based on the filtered depth map to determine the edges, thereby obtaining the depth edge information; The point cloud image is denoised using a random sampling consensus algorithm to obtain a filtered point cloud image, and principal component analysis is performed on the filtered point cloud image to obtain the point cloud edge set.

3. The method for processing a magnetic material substrate using spray coating protective tooling according to claim 2, characterized in that, The grayscale edge information, the depth edge information, and the point cloud edge set are fused to obtain a three-dimensional fused feature point cloud, including: The grayscale edge information and the depth edge information are mapped to three-dimensional space using camera intrinsic parameters and camera extrinsic parameters, respectively, to obtain three-dimensional grayscale edge point cloud and three-dimensional depth edge point cloud; The three-dimensional grayscale edge point cloud, the three-dimensional depth edge point cloud, and the point cloud edge set are weighted and fused to obtain a three-dimensional fused feature point cloud.

4. The method for processing a magnetic material substrate using spray coating protective tooling according to claim 2, characterized in that, Using camera intrinsic and extrinsic parameters, the grayscale edge information and the depth edge information are mapped to three-dimensional space respectively to obtain three-dimensional grayscale edge point clouds and three-dimensional depth edge point clouds, including: Extract edge pixels from the grayscale edge information and the depth edge information respectively to obtain grayscale pixel coordinates and depth pixel coordinates; The grayscale pixel coordinates and the depth pixel coordinates are transformed to the normalized plane in the camera coordinate system using the camera intrinsic parameters, respectively, to obtain the grayscale pixel camera coordinates and the depth pixel camera coordinates. Based on the depth value corresponding to the pixel camera coordinates in the camera coordinate system, the grayscale pixel camera coordinates and the depth pixel camera coordinates are converted into pixel coordinates in the three-dimensional camera coordinate system to obtain the grayscale pixel three-dimensional camera coordinates and the depth pixel three-dimensional camera coordinates. The grayscale pixel 3D camera coordinates and the depth pixel 3D camera coordinates are transformed to the world coordinate system using the camera extrinsic parameters to obtain the 3D grayscale edge point cloud and the 3D depth edge point cloud.

5. The method for processing a magnetic material substrate using spray coating protective tooling according to claim 1, characterized in that, Based on the 3D fused feature point cloud, region connectivity processing is performed to obtain the 3D spatial distribution features of the target region, including: The three-dimensional fused feature point cloud is spatially meshed to obtain a three-dimensional voxel structure; Extract the local maxima of the gradient in the three-dimensional voxel structure to obtain the preliminary three-dimensional edge; The gradient magnitude of the preliminary three-dimensional edge is filtered using a dual threshold method to obtain the first target edge and the second target edge; The first target edge and the second target edge are connected by a region growing algorithm to obtain the three-dimensional edge of the target; Based on connected component analysis, the independent edge segments of the target's three-dimensional edge are identified to obtain the three-dimensional spatial distribution characteristics.

6. The method for processing a magnetic material substrate using a spray-coating protective fixture according to claim 5, characterized in that, The reconstructed surface model is meshed to generate a meshed damage model, including: The vertex set of the reconstructed surface model is subjected to Delaunay triangulation to obtain a preliminary triangular mesh; Calculate the quadratic error matrix for each vertex of the initial triangular mesh, and construct an edge priority queue sorted by shrinkage cost based on the quadratic error matrix; The edge with the minimum shrinkage cost is selected from the edge priority queue for shrinkage operation, the mesh topology is updated, and the target triangular mesh is obtained. The target triangular mesh is filtered to obtain the meshed damage model.

7. The method for processing a magnetic material substrate using spray coating protective tooling according to claim 1, characterized in that, Converting the two-dimensional flattened model into a laser cutting file includes: Based on the two-dimensional flattened model, the boundary is extracted, the outer contour is separated, and the outer contour separation file is obtained. The external contour separation file is converted to a different format to obtain the laser cutting file.

8. A spray coating protective tooling device for a magnetic material substrate, characterized in that, The device includes: The image acquisition module is used to acquire grayscale images, depth maps, and point cloud images of the target area on the magnetic material substrate. The image processing module is used to perform filtering and feature extraction on the grayscale image, the depth image and the point cloud image respectively to obtain grayscale edge information, depth edge information and point cloud edge set; The feature fusion module is used to fuse the grayscale edge information, the depth edge information, and the point cloud edge set to obtain a three-dimensional fused feature point cloud; The feature extraction module is used to perform region connectivity processing based on the three-dimensional fused feature point cloud to extract the three-dimensional spatial distribution features of the target region; The model building module is used to reconstruct the surface based on the three-dimensional spatial distribution features to obtain a reconstructed surface model of the target area. The model processing module is used to perform meshing processing on the reconstructed surface model to generate a meshed damage model; The model flattening module is used to perform two-dimensional unfolding of the meshed damage model to obtain a two-dimensional flattened model. The laser cutting module is used to convert the two-dimensional flattened model into a laser cutting file, and to perform laser cutting on the magnetic material workpiece according to the laser cutting file to obtain the spraying protective fixture for the target area. The step of reconstructing the surface based on the three-dimensional spatial distribution features to obtain a reconstructed surface model of the target region includes: The gradient magnitude of each voxel in the three-dimensional spatial distribution features is obtained, and the gradient magnitude is linearly normalized to obtain the normalized magnitude. The normalized amplitude is then subjected to a nonlinear transformation to obtain a nonlinear amplitude. Isosurfaces are extracted based on the nonlinear amplitude, and initial triangular patches are generated based on the isosurfaces; The initial triangular facet is optimized using a quality error metric to obtain the target triangular facet. The target triangular facets are connected and converted into a target format to obtain the reconstructed surface model; The step of unfolding the meshed damage model into a two-dimensional flattened model includes: Construct a graphical Laplacian matrix based on the meshed damage model; Calculate the eigenvalues ​​and eigenvectors of the graph Laplacian matrix; Select a target number of non-trivial feature values ​​and non-trivial feature vectors based on the feature values ​​and feature vectors, respectively. A two-dimensional embedding is constructed based on the non-trivial eigenvalues ​​and the non-trivial eigenvectors to obtain a two-dimensional point set; The two-dimensional point set is optimized by conformal mapping and damage constraint terms are added to obtain the two-dimensional flattened model. The damage constraint terms are added as follows: In the above formula, u Represents a two-dimensional point set, u i Indicates the first i The two-dimensional points corresponding to the vertices of the target triangular mesh. u j Indicates the first j The two-dimensional points corresponding to the vertices of the target triangular mesh. m and n Indicates the penalty coefficient. Indicates the minimum allowable distance. d i and d j They represent the first i The vertices of the target triangular mesh and the first j Damage values ​​of the vertices of the target triangular mesh. τ This represents the threshold for damage similarity.

Images