Method for processing three-dimensional point cloud data of complex steel structure, reconstruction method and electronic device

Abstract

The embodiment of the application discloses a kind of complex steel structure three-dimensional point cloud data processing method, reconstruction method, electronic equipment.Therein, three-dimensional point cloud data processing method includes: obtaining the three-dimensional point cloud data of complex steel structure;According to the position relationship between adjacent points, the three-dimensional point cloud data is decomposed into multiple plane clusters;For each described plane cluster, uv surface projection is carried out based on local coordinate system, wherein the local coordinate system includes u axis, v axis and w axis, w axis is perpendicular to the plane cluster, u axis, v axis and w axis are perpendicular to each other;For each described plane cluster, according to the projection point density distribution of the uv surface projection to generate binary image.The embodiment of the application can effectively reduce the amount of calculation.

Description

Technical Field

[0001] This application relates to the fields of complex steel structures and 3D reconstruction, and more specifically, to a method for processing 3D point cloud data of complex steel structures, a reconstruction method, and an electronic device. Background Technology

[0002] In the current fields of industrial manufacturing and construction engineering, the digital inspection and reconstruction of complex steel structures (such as large bridge nodes and ship cabins) has become a critical requirement. Among them, reverse reconstruction technology based on 3D laser scanning is the core means to obtain accurate three-dimensional models, but existing processing methods still mainly rely on manual or traditional algorithms, which are inefficient and difficult to handle complex structures. Summary of the Invention

[0003] This application provides a method for processing and reconstructing three-dimensional point cloud data of complex steel structures, as well as an electronic device, to effectively reduce the amount of computation.

[0004] According to a first aspect of the embodiments of this application, a method for processing three-dimensional point cloud data of complex steel structures is provided, including: Acquire 3D point cloud data of complex steel structures; Based on the positional relationship between adjacent points, the three-dimensional point cloud data is decomposed into multiple planar clusters; For each of the plane clusters, a uv plane projection is performed based on a local coordinate system, wherein the local coordinate system includes a u-axis, a v-axis, and a w-axis, the w-axis is perpendicular to the plane cluster, and the u-axis, v-axis, and w-axis are mutually perpendicular; For each of the plane clusters, a binary image is generated based on the density distribution of the projection points of the uv surface projection.

[0005] According to a second aspect of the embodiments of this application, a method for reconstructing a three-dimensional point cloud of a complex steel structure is provided, comprising: Binary images of multiple planar clusters are obtained using the method provided in the first aspect; For each binary image, perform the following processing: Extract the closed contours and determine the hierarchical relationship between them; Based on the hierarchical relationship between the closed contours, contour merging and cutting operations are performed to obtain 2D patches; Based on the 2D patches, the local coordinate system is mapped to the global world coordinate system so as to splice and generate a triangular network model in the global world coordinate system.

[0006] A third aspect of the present application provides an electronic device including a memory and a processor. The memory stores computer instructions, and the processor invokes and executes the computer instructions to implement the method described in the first or second aspect.

[0007] The method provided in this application combines local coordinate system projection and binarized images to transform complex 3D point cloud data into a relatively mature 2D image processing problem. Compared with traditional techniques that directly perform complex Boolean operations in three-dimensional space, resulting in an exponential increase in computational load, the binary image can carry feature data, thereby effectively reducing the computational load of subsequent data processing. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the hardware and data flow of a complex steel structure reverse reconstruction system according to an embodiment of this application; Figure 2 This is a flowchart illustrating a method for processing three-dimensional point cloud data of a complex steel structure according to an embodiment of this application. Figure 3 This is a flowchart illustrating a method for reconstructing a three-dimensional point cloud of a complex steel structure according to an embodiment of this application. Figure 4 This is a flowchart of a steel structure reverse reconstruction method based on planar cluster projection and odd-even hierarchy mapping according to an embodiment of this application; Figure 5 This is a schematic diagram of a region-growing planar cluster segmentation process according to an embodiment of this application; Figure 6 This is a schematic diagram of an odd-even layer plane merging and cutting principle according to an embodiment of this application; Figure 7 This is a schematic diagram of a regional growth planar cluster according to an embodiment of this application; Figure 8 This is a schematic diagram of an exemplary planar cluster segmentation result according to an embodiment of this application; Figure 9 This is a schematic diagram of the final reverse modeling result according to an embodiment of this application. Detailed Implementation

[0009] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0010] It should be understood that "multiple" as mentioned herein refers to two or more. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, to facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first," "second," etc., are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or execution order, and the terms "first," "second," etc., do not necessarily imply differentness. Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0011] like Figure 1 The diagram shown is a schematic of the hardware and data flow of the complex steel structure reverse reconstruction system according to an embodiment of this application. It mainly includes an industrial robot, a laser line scanning camera, a complex steel structure, and a computing processing terminal.

[0012] Data acquisition unit: Includes a high-precision line scan camera 1. This camera projects laser stripes onto the target surface and, in conjunction with the relative displacement of the motion mechanism, captures the original 3D point cloud sequence P of the target surface in real time. The data contains high-precision spatial coordinates. It also includes reflection depth information, which can fully record the surface geometric features and texture details of the steel structure.

[0013] Motion and Execution System: This system comprises an industrial robot (2) and the complex steel structure under test (3). The industrial robot, acting as a high-precision motion carrier, carries a line-scanning camera along a preset trajectory or drives the end effector of the robotic arm to adjust its posture, ensuring a constant working distance and scanning angle between the camera's optical axis and the steel structure surface. The complex steel structure under test, as the core object being measured, has multiple curved surfaces, welds, and irregular features. During the scanning process, the robot guides the camera to perform full-coverage acquisition of key components such as the main wing plate, secondary wing plate, and base plate of the steel structure, ensuring the continuity and integrity of the point cloud data and providing a realistic physical benchmark for subsequent modeling.

[0014] The computing processing end includes a high-performance industrial control computer. Internally, it integrates a multi-frame point cloud stitching and registration module, a normal vector estimation unit, a region growing and segmentation algorithm library, a local coordinate system builder, and a 2D image processing engine. This end is responsible for executing the entire algorithm process, from multi-source point cloud denoising, fusion and resampling, to complex steel structure topology reconstruction and final mesh generation. It is specifically optimized for the elongated stripe distribution characteristics unique to line scan data, ultimately outputting a high-precision digital model file.

[0015] Figure 2 This is a flowchart illustrating a method for processing three-dimensional point cloud data of a complex steel structure according to an embodiment of this application. (Refer to...) Figure 2 The method includes the following processing steps.

[0016] 20: Obtain 3D point cloud data of complex steel structures. For example, the 3D point cloud data can be obtained using... Figure 1 The data was collected using the method provided in the illustrated embodiment.

[0017] 22: Based on the positional relationship between adjacent points, the three-dimensional point cloud data is decomposed into multiple planar clusters.

[0018] 24: For each of the plane clusters, perform uv-plane projection based on a local coordinate system. The local coordinate system includes a u-axis, a v-axis, and a w-axis, where the w-axis is perpendicular to the plane cluster, and the u-axis, v-axis, and w-axis are mutually perpendicular. uv-plane projection refers to projection along a direction perpendicular to the uv-plane.

[0019] 26: For each of the plane clusters, a binary image is generated based on the density distribution of the projection points of the uv surface projection.

[0020] The method provided in this application combines local coordinate system projection and binarized images to transform complex 3D point cloud data into a relatively mature 2D image processing problem. Compared with traditional techniques that directly perform complex Boolean operations in three-dimensional space, resulting in an exponential increase in computational load, the binary image can carry feature data, thereby effectively reducing the computational load of subsequent data processing.

[0021] Optionally, in one implementation of this embodiment, the positional relationship includes the angle between normal vectors and the distance, wherein the angle between the normal vectors of any two points within the same planar cluster is less than an angle threshold and the distance is less than a distance threshold. In this embodiment, the planar cluster refers to a cluster of approximately planar elements, which can be obtained through a region growing algorithm.

[0022] For example, for any point in a point cloud, the covariance matrix of the points in its neighborhood (four-neighborhood or eight-neighborhood) is calculated and eigenvalue decomposition is performed to determine the principal component directions, thereby obtaining the normal vector of the point in the local space. Specifically, after determining the principal component directions, the covariance matrix is ​​constructed, eigenvalue decomposition is performed, and the normal vector (i.e., the eigenvector corresponding to the smallest eigenvalue) is selected. This provides a foundation for subsequently distinguishing different planar clusters and effectively solves the problem of feature extraction failure due to noise in traditional methods.

[0023] Optionally, in one implementation of this embodiment, between processing steps 22 and 24, the method further includes: discarding planar clusters whose thickness is outside a set range based on the thickness of each planar cluster in the w-axis direction. Specifically, principal component analysis (PCA) can be used to extract their principal directions and construct a local coordinate system. The w-axis is perpendicular to the plane, and the thickness distribution of the cluster in the w-axis direction is calculated. Abnormal clusters that are too thin (e.g., 10% of all thicknesses are used as the lower limit of the thickness threshold to determine whether they are too thin) or too thick (e.g., 90% of all thicknesses are used as the upper limit of the thickness threshold to determine whether they are too thick) are eliminated, and finally the effective planar clusters are left, thus solving the problem of boundary ambiguity when multiple planes intersect.

[0024] Optionally, in one implementation of this embodiment, a binary image is generated based on the density distribution of projection points of the UV surface projection: The binary image is generated using histogram statistics or an adaptive thresholding method based on the density distribution of projection points, wherein positions with illuminated points are marked as 1, and positions without illuminated points are marked as 0. This implementation transforms the originally complex 3D geometric relationships into an intuitive 2D pixel matrix, achieving dimensionality reduction of the problem / features.

[0025] Optionally, in one implementation of this embodiment, after processing 26, the method further includes: performing dilation and erosion operations on the binary image to remove noise or burrs from the binary image. Using this implementation, after the opening operation of "dilation + erosion," a clean binary image with clear contours and smooth boundaries is obtained, laying a solid foundation for subsequent contour extraction.

[0026] This application also provides a method for reconstructing a three-dimensional point cloud of a complex steel structure. Exemplarily, in the three-dimensional point cloud reconstruction method, a method can be employed... Figure 1 The method described in the illustrated embodiment involves removing planar clusters whose thickness is outside a set range. After obtaining the binary image, an inverse projection transformation is performed to map the coordinates back to the global world coordinate system. Then, traditional computational methods can be used for 3D point cloud reconstruction. This approach effectively removes point cloud data related to abnormal clusters, reducing computational load while resolving the boundary blurring problem when multiple planes intersect.

[0027] This application also provides a method for reconstructing a three-dimensional point cloud of a complex steel structure. Exemplarily, in the three-dimensional point cloud reconstruction method, a method can be employed... Figure 1 The method described in the illustrated embodiment involves performing an inverse projection transformation on the obtained binary image to map the coordinates back to the global world coordinate system. Afterwards, traditional computational methods can be used for 3D point cloud reconstruction. This approach effectively removes point cloud data related to anomalous clusters, reducing computational load while eliminating point cloud data that causes unclear boundaries and contours, thus improving data quality.

[0028] Figure 3 This is a flowchart illustrating a method for reconstructing a three-dimensional point cloud of a complex steel structure according to an embodiment of this application. (Refer to...) Figure 3 The method includes the following processing steps.

[0029] 30: Obtain the binary image. Please refer to the following for specific methods. Figure 2 The illustrated embodiments or their implementations.

[0030] 32: For each binary image, extract the closed contours and determine the hierarchical relationship between the closed contours.

[0031] 34: Perform contour merging and cutting operations based on the hierarchical relationship between the closed contours to obtain 2D patches.

[0032] 36: Based on the 2D patch, perform the mapping from the local coordinate system to the global world coordinate system, so as to splice and generate a triangular network model in the global world coordinate system.

[0033] The method provided in this embodiment automatically distinguishes between solids and internal cavities of complex steel structures by using hierarchical contour mapping. It reduces the computationally intensive 3D spatial Boolean operations to efficient layer-by-layer 2D operations, solving the problems of exponential growth in computational load and easy occurrence of topological errors when dealing with complex internal cavity structures using traditional methods. This significantly improves the efficiency of automated reconstruction and model integrity of complex steel structures.

[0034] Optionally, in one implementation of this embodiment, determining the hierarchical relationship between closed contours includes: determining whether a closed contour belongs to an even-numbered layer or an odd-numbered layer based on the nesting relationship between the closed contours, wherein the outermost closed contour is an even-numbered layer. The step of merging and cutting contours based on the hierarchical relationship between the closed contours to obtain 2D patches includes: merging all even-numbered layer closed contours to fuse scattered solid regions into complete solid patches; and, based on the result of the merging operation, cutting all odd-numbered layer closed contours to subtract the odd-numbered layer closed contours from the solid patches using them as masks. This implementation follows the mapping rule of "even-numbered layers as solids, odd-numbered layers as holes." The essence of cutting odd-numbered layers can be understood as "sculpting" by cutting waste material, while even-numbered layers can be understood as "sculpting" by splicing and shaping. Even-numbered layers create solids, while odd-numbered layers cut away useless content.

[0035] Optionally, in one implementation of this embodiment, the mapping from the local coordinate system to the global world coordinate system based on the 2D patches includes: stacking the 2D patches corresponding to the plurality of planar clusters along the w-axis, and mapping the coordinates back to the global world coordinate system through inverse projection transformation. This yields hole-removed and noise-reduced entity point cloud data.

[0036] Figure 4 This is a flowchart of a steel structure reverse reconstruction method based on planar cluster projection and odd-even hierarchy mapping according to an embodiment of this application. It mainly includes the following processing steps: acquiring original point cloud data P, normal vector estimation module, planar cluster segmentation based on region growing, 3D projection to 2D, 2D image processing, contour extraction and odd-even hierarchy construction, layer-by-layer calculation and coordinate inverse transformation, model synthesis and output. Figure 5 This is a schematic diagram of a region growing planar cluster segmentation process according to an embodiment of this application. Figure 6 This is a schematic diagram of the principle of merging and cutting odd and even layer planes according to an embodiment of this application. Figure 7 This is a schematic diagram of a regional growth planar cluster according to an embodiment of this application.

[0037] The following is combined with Figures 4-7 A method for reverse reconstruction of steel structures based on planar cluster projection and odd-even hierarchy mapping according to an embodiment of this application will be described.

[0038] (1) Acquisition of raw point cloud and estimation of normal vector: The target steel structure is scanned from all directions using a laser line scanning camera to acquire raw point cloud data containing millions of points. To eliminate the influence of outliers and random noise generated during the scanning process, the system first calculates the normal vector Ni in the local neighborhood of all points Pi. Then, based on the consistency of the normal vector or other geometric features, outliers and noise points are identified and removed to obtain clean point cloud data. By calculating the covariance matrix of points in the neighborhood of a point (4-neighborhood, which contains four directly adjacent point clouds in the top, bottom, left, and right directions; 8-neighborhood is the four point clouds containing the 4-neighborhood, plus the four diagonal point clouds in the top left, top right, bottom left, and bottom right directions) and performing eigenvalue decomposition, the principal component direction is determined, thereby obtaining the normal vector of the point in the local space. This step is the basis for distinguishing different planar clusters and effectively solves the problem of feature extraction failure caused by noise in traditional methods. (2) Planar cluster segmentation based on region growing: The specific flowchart is as follows Figure 3 As shown, after obtaining the normal vector, the system uses a "region growing" algorithm to segment the point cloud into several independent planar clusters. For example, algorithms such as Basic Region Growing and K-Means Initialization RG can be used. In practical applications, Basic Region Growing is used to select a seed point (e.g., a corner point within the entire point cloud region, which serves as the starting point for growing the entire planar cluster), and its 8-neighborhood point clouds are checked. If a threshold condition is met, the region is added. This process is repeated until no new point clouds are added.

[0039] Set angle threshold (e.g., 15 degrees) and a distance threshold d (e.g., 5 mm). Traverse the seed points in the point cloud; if the angle between the normal vectors of adjacent points is less than... If the distance is less than d, the point is grouped into the same cluster. Through iterative growth, the complex global point cloud is decomposed into several approximate planar local clusters. For a detailed schematic diagram of planar cluster segmentation based on region growing, please refer to... Figure 5 For each planar cluster, the system further performs principal component analysis (PCA) to extract its principal directions and construct a local coordinate system. The w-axis is perpendicular to the plane, and the thickness distribution of the cluster in the w-axis direction is calculated. Abnormal clusters that are too thin or too thick are eliminated, and finally the effective planar clusters are left, which solves the problem of boundary ambiguity when multiple planes intersect.

[0040] Example segmentation results are as follows Figure 8 As shown, different colors represent different planar clusters.

[0041] (3) 3D to 2D projection and binary image generation: The 3D points within each segmented planar cluster are projected onto the uv plane based on their local coordinate system to form a 2D point set. Subsequently, based on the density distribution of the projected points, an initial binary image is generated using histogram statistics or an adaptive thresholding method. In this image, the location of a lit point is marked as "1" (representing solid material), and the location of a non-lit point is marked as "0" (representing background or gaps). At this point, the originally complex 3D geometric relationship is transformed into an intuitive 2D pixel matrix, achieving dimensionality reduction of the problem. (4) Morphological processing and noise removal: For the small noise points (such as isolated points caused by scanning errors) or edge burrs that still exist in the binary image generated by projection, the system performs standard morphological operations. First, a "dilation" operation is performed to fill the small holes in the image and connect the broken edges; then a "erosion" operation is performed to restore the original size of the object and remove the excess noise caused by dilation. After the opening operation of "dilation + erosion", a clean binary image with clear contours and smooth boundaries is obtained, laying a solid foundation for subsequent contour extraction. (5) Contour extraction and odd-even hierarchical structure construction The Canny operator is used to extract all closed contours from a binary image. The system automatically constructs a hierarchical tree structure based on the nesting relationship of the contours: the outermost contour is defined as layer 0 (solid), the first nested contour is defined as layer 1 (hole), the second nested contour is defined as layer 2 (solid), and so on. It follows the mapping rule of "even-numbered layers are solids, odd-numbered layers are holes." The essence of cutting at odd-numbered layers can be understood as "sculpting" by cutting waste material, while even-numbered layers can be understood as "sculpting" by piecing together and shaping. Even-numbered layers create solids, while odd-numbered layers cut out useless content.

[0042] (6) Layer-by-layer 2D Boolean operations and inverse coordinate transformation The system converts the above hierarchical structure into layer-by-layer 2D Boolean operation instructions. For a detailed flowchart of the odd-number splitting and even-number merging layer processing, please refer to [link / details]. Figure 6 .

[0043] Even-numbered layer processing: Perform a "Fuse / Union" operation on the contours of all even-numbered layers to merge scattered solid regions into complete solid patches. The formula is expressed as:

[0044] This represents the information for even-numbered layer solid faces, where 2k indicates an even-numbered layer, 2k+1 indicates an odd-numbered layer, and U stands for sum. Represents even-numbered slice information. This represents information from odd-numbered layers.

[0045] Odd-numbered layer processing: Perform a "Cut / Subtract" operation on the contours of all odd-numbered layers, using them as a mask to subtract from the corresponding even-numbered layer entities. The formula is as follows:

[0046] in, For even-numbered layer entity information, Information on the surface features of solid regions with odd-numbered contour levels. This represents the entity patch information after processing all odd and even layers.

[0047] After completing all levels of 2D Boolean operations, the system stacks these processed 2D patches along the w-axis of the local coordinate system (i.e., the original 3D height direction) and remaps the coordinates back to the global world coordinate system through inverse projection transformation.

[0048]

[0049] in, These are the 2D patch coordinates in the global world coordinate system. for The coordinates of all 2D faces of the entities in the local coordinate system. , These are the rotation and translation matrices for transforming 2D surfaces to the global world coordinate system.

[0050] (7) Model Synthesis and Output: Finally, the system stitches together the processing results of each layer into a continuous triangular mesh model. Due to the dimensionality reduction processing, the model not only retains the geometric features of the original steel structure, but also completely eliminates topological errors such as self-intersection and normal reversal commonly found in traditional 3D Boolean operations. The final output model can be directly imported into CAD software for subsequent finite element analysis or CNC machining programming, realizing high-precision and high-efficiency automated reconstruction of complex steel structures.

[0051] For example, the result of reverse modeling is as follows: Figure 9 As shown.

[0052] By employing the embodiments of this application, the problems of difficult multi-plane intersection recognition, weak noise resistance, and low computational efficiency of complex steel structures are effectively solved by reducing 3D point clouds to 2D hierarchical contours and using parity mapping logic. This method utilizes morphological processing to accurately filter noise and eliminate model fractures, and achieves high-precision modeling of internal cavity structures through an automated "solid merging and hole cutting" mechanism. Compared with traditional full 3D computation, this solution significantly reduces computational complexity and greatly improves reconstruction speed and accuracy. It not only avoids manual intervention but also meets the urgent need for rapid and efficient digital reconstruction of complex steel structures in engineering sites. This application also provides an electronic device, which includes a memory and a processor. The memory stores computer instructions, and the processor invokes and executes the computer instructions to implement the method embodiments described above. For a detailed explanation of the implementation process, please refer to the preceding text; it will not be repeated here.

[0053] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0054] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0055] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0056] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The usable medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital versatile discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)). It is worth noting that the computer-readable storage medium mentioned in the embodiments of this application can be a non-volatile storage medium; in other words, it can be a non-transient storage medium. It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in the embodiments of this application are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the scene data of the current frame in the 3D virtual scene involved in the embodiments of this application, the client's device information, and the scene interaction information are all obtained with full authorization.

[0057] The above are merely preferred embodiments of this application. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for processing three-dimensional point cloud data of complex steel structures, characterized in that, The method includes: Acquire 3D point cloud data of complex steel structures; Based on the positional relationship between adjacent points, the three-dimensional point cloud data is decomposed into multiple planar clusters; For each of the plane clusters, a uv plane projection is performed based on a local coordinate system, wherein the local coordinate system includes a u-axis, a v-axis, and a w-axis, the w-axis is perpendicular to the plane cluster, and the u-axis, v-axis, and w-axis are mutually perpendicular; For each of the plane clusters, a binary image is generated based on the density distribution of the projection points of the uv surface projection.

2. The method according to claim 1, characterized in that, The positional relationship includes the angle between normal vectors and the distance. The angle between the normal vectors of any two points within the same plane cluster is less than an angle threshold and the distance is less than a distance threshold.

3. The method according to claim 2, characterized in that, The normal vector of a single point is determined in the following way: Calculate the covariance matrix of points in the neighborhood of the single point and perform eigenvalue decomposition to determine the principal component direction, thereby obtaining the normal vector of the single point in the local space.

4. The method according to claim 1, characterized in that, The method further includes: After obtaining the plurality of planar clusters, planar clusters whose thickness is not within the set range are discarded according to the thickness of each planar cluster in the w-axis direction.

5. The method according to claim 1, characterized in that, The step of generating a binary image based on the projection point density distribution of the UV surface projection includes: Based on the density distribution of the projection points, the binary image is generated using histogram statistics or an adaptive thresholding method, wherein the position with a light point is marked as 1, and the position without a light point is marked as 0.

6. The method according to claim 1, characterized in that, The method further includes: Dilation and erosion operations are performed on the binary image to remove noise or burrs.

7. A method for reconstructing three-dimensional point clouds of complex steel structures, characterized in that, The method includes: Binary images of multiple planar clusters are obtained using the method described in any one of claims 1-6; For each binary image, perform the following processing: Extract the closed contours and determine the hierarchical relationship between them; Based on the hierarchical relationship between the closed contours, contour merging and cutting operations are performed to obtain 2D patches; Based on the 2D patches, the local coordinate system is mapped to the global world coordinate system so as to splice and generate a triangular network model in the global world coordinate system.

8. The method according to claim 7, characterized in that, The determination of the hierarchical relationship between closed contours includes: The nesting relationship between closed contours determines whether a closed contour belongs to an even or odd layer, where the outermost closed contour is an even layer. The step of merging and cutting contours based on the hierarchical relationship between the closed contours to obtain 2D patches includes: Merge the closed contours of all even-numbered layers to combine scattered solid regions into a complete solid patch. Based on the result of the merging operation, a cutting operation is performed on the closed contours of all odd-numbered layers to subtract the closed contours of the odd-numbered layers from the solid patches as masks.

9. The method according to claim 7, characterized in that, The mapping from the local coordinate system to the global world coordinate system based on the 2D patch includes: The 2D patches corresponding to the multiple planar clusters are stacked along the w-axis, and the coordinates are mapped back to the global world coordinate system through inverse projection transformation.

10. An electronic device, characterized in that, The electronic device includes: Memory, used to store computer instructions; A processor for invoking and executing the computer instructions to implement the method as described in any one of claims 1-9.

Images