Complex wiring preprocessing method and system based on 3D point cloud measurement and registration technology

Abstract

The invention provides a complex wiring preprocessing method and system based on a 3D point cloud measurement and registration technology. The method belongs to the field of pipeline routing design in a complex environment. The method comprises the following steps: acquiring a point cloud image of a target routing position through a binocular camera to obtain multiple sections of point cloud data; cutting repeated point cloud data in the multiple segments of point cloud data; voxel grid filtering is carried out on the cut point cloud data; performing normal estimation on the point cloud data after voxel grid filtering; extracting a local feature of each point to obtain a geometrical relationship between each point and a neighborhood point of the point; calculating an initial transformation matrix between the point clouds by using an SAC-IA algorithm; performing fine registration on the two complete point clouds by using an ICP (Inductively Coupled Plasma) point cloud registration algorithm to obtain a final point cloud image of the target routing position; and outputting the wiring size of the target wiring position according to the final point cloud image for wiring design. According to the method, the point cloud image can be generated in a complex environment, so that the wiring size is obtained, and a user can conveniently perform pipeline wiring design.

Description

Technical Field

[0001] The present application relates to the field of pipeline routing design in complex environments, and more specifically, to a complex routing preprocessing method and system based on 3D point cloud measurement and registration technology. Background Art

[0002] When dealing with the rigid steel structure configuration at the bottom of the train and the intricate wiring of the rails and railways, engineers face the great challenge of accurately measuring and surveying the steel pipes in the line. These steel pipes often exist in a cross-shaped, dense and narrow space, which greatly increases the difficulty of accurately measuring the target distance between points in the existing pipelines.

[0003] Currently, engineers mainly rely on physical tools such as tape measures or more advanced infrared distance sensors to assist in surveying and mapping. However, tape measures not only have large measurement errors, but also lack the ability to avoid obstacles, making it difficult to cope with complex environments such as crossing, dense, and narrow in routing measurements. In contrast, although infrared distance sensors can alleviate these engineering problems to a certain extent, using infrared light to penetrate the wiring layout of dense and narrow areas to achieve surveying and distance measurement, due to the unidirectional propagation characteristics of light, when the path is fully covered by obstacles, the infrared light is reflected in advance before reaching the target point. Therefore, infrared distance sensors are still insufficient in solving the problem of cross obstacles. Specifically, it cannot effectively measure the actual distance between the surveying point and the target point on a fully obstructed surveying path.

[0004] The wiring of the train bottom and rails is often complicated. In most cases, curved wiring and large-angle and multi-angle wiring on the plane are required. Traditional methods can only complete the mapping between straight line points, which is powerless for long-distance surveys, especially when the wiring becomes tortuous. Summary of the invention

[0005] The purpose of this application is to provide a complex routing preprocessing method and system based on 3D point cloud measurement and registration technology, which can generate point cloud images in complex environments to obtain wiring dimensions to facilitate users to design pipeline routing.

[0006] This application is implemented as follows:

[0007] In a first aspect, the present application provides a complex routing preprocessing method based on 3D point cloud measurement and registration technology, comprising:

[0008] The target routing position is photographed in segments by using a binocular depth camera, and each shot generates a segment of point cloud data to obtain multiple segments of point cloud data;

[0009] Crop repeated point cloud data in multiple point cloud data segments;

[0010] Perform voxel grid filtering on the cropped point cloud data;

[0011] The normal line of the point cloud data after voxel grid filtering is estimated to obtain the surface normal line feature of each point in the point cloud;

[0012] Extract the local normal features of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points;

[0013] Based on the geometric relationship between each point and its neighboring points, the SAC-IA algorithm is used to calculate the initial transformation matrix between the point clouds. Based on the initial transformation matrix between the point clouds, the overlapping point clouds of the two complete point clouds are used to perform preliminary registration, so that the two point clouds are initially aligned globally.

[0014] Use the ICP point cloud registration algorithm to accurately register the two complete point clouds to obtain the final point cloud image of the target alignment position;

[0015] The wiring dimensions of the target routing position are output according to the final point cloud image for routing design.

[0016] Based on the first aspect, the step of cutting out repeated point cloud data in the multi-segment point cloud data includes:

[0017] Manually select the same point cloud data and crop it.

[0018] Based on the first aspect, the step of performing voxel grid filtering on the cropped point cloud data includes:

[0019] Use the key function setLeafSize(0.001,0.001,0.001) in the voxel grid filter to smooth or lose details or features smaller than 1 mm.

[0020] Based on the first aspect, the step of performing normal estimation on the point cloud data after the voxel grid filtering to obtain the surface normal feature of each point in the point cloud includes:

[0021] The key function setKSearch(20) is used to set the number of nearest neighbor points considered when calculating the normal to 20. Each point estimates its normal based on the 20 nearest neighbor points around it.

[0022] Based on the first aspect, the steps of extracting the local normal feature of each point and calculating the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points include:

[0023] Use the key function setKSearch(30) to set the number of nearest neighbor points considered when calculating features to 30. Each point calculates its features based on the 30 nearest neighbor points around it.

[0024] Based on the first aspect, the step of using the SAC-IA algorithm to calculate the initial transformation matrix between point clouds includes:

[0025] Use the key function setMaxCorrespondenceDistance(0.01) to set the maximum distance threshold when determining corresponding point pairs. The distance between two points must be within 0.01 meters to be considered a valid corresponding point.

[0026] Based on the first aspect, the steps of using the ICP point cloud registration algorithm to accurately register two complete point clouds to obtain the final point cloud image of the target alignment position include:

[0027] Set the parameters of the ICP point cloud registration algorithm to: setMaximumIterations(100), setTransformationEpsilo(1e-20), setEuclideanFitnessEpsilon(1e-20).

[0028] In a second aspect, the present invention provides a complex routing preprocessing system based on 3D point cloud measurement and registration technology, comprising:

[0029] The acquisition module is used to shoot the target routing position in segments through a binocular depth camera, and each shooting generates a segment of point cloud data to obtain multiple segments of point cloud data;

[0030] A cropping module is used to crop repeated point cloud data in multiple point cloud data segments;

[0031] Downsampling module, used to perform voxel grid filtering on the cropped point cloud data;

[0032] The normal estimation module is used to estimate the normal of the point cloud data after voxel grid filtering to obtain the surface normal features of each point in the point cloud;

[0033] A feature extraction module is used to extract the local normal feature of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points;

[0034] The coarse registration module is used to calculate the initial transformation matrix between point clouds based on the geometric relationship between each point and its neighboring points using the SAC-IA algorithm. Based on the initial transformation matrix between point clouds, the initial registration is performed between two complete point clouds using their overlapping point clouds, so that the two point clouds are initially aligned globally.

[0035] The precise registration module is used to precisely register two complete point clouds using the ICP point cloud registration algorithm to obtain the final point cloud image of the target alignment position;

[0036] The routing design module is used to output the wiring dimensions of the target routing position according to the final point cloud image for routing design.

[0037] In a third aspect, the present invention provides an electronic device, comprising:

[0038] A memory for storing one or more programs;

[0039] processor;

[0040] When the one or more programs are executed by the processor, the above method is implemented.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the above method when executed by a processor.

[0042] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0043] The tape measure or infrared distance detector used in the prior art measures the required wiring dimensions of the steel pipe, but in the face of cross-sectional, dense and narrow areas, it is usually impossible to measure because of obstruction during measurement. The present invention uses a binocular depth camera to obtain a point cloud image of the target routing position. When encountering an obstructed situation, the position and angle of the binocular depth camera can be changed to perform segmented shooting to obtain multiple point cloud data. The acquisition operation is convenient and does not require much physical work. It completely solves the routing obstacle problem, ignores obstacles and the narrow space, and has high measurement accuracy. For the situation where the target routing position has long-distance and multi-angle routing bends, the present invention can also obtain multi-segment point cloud data by segmented shooting of the target routing position, and obtain the image of the entire target routing position through subsequent registration of the multi-segment point cloud data, so that the user can design the routing according to the image of the target routing position.

[0044] By cropping the repeated point cloud data, the amount of calculation of the subsequent stitching algorithm is reduced and the stitching accuracy is improved. By performing voxel grid filtering on the cropped point cloud data, the amount of calculation is further reduced and the running speed is accelerated. By setting the parameter setKSearch(20) for normal estimation, the accuracy and stability of normal estimation are ensured. By setting setKSearch(30) for feature extraction, the detailed geometric structure of the point cloud can be captured while avoiding the introduction of irrelevant points. By setting the parameter setMaxCorrespondenceDistance(0.01) for coarse alignment, the accuracy and stability of the alignment result are ensured. By setting the parameters of the ICP point cloud alignment algorithm for fine alignment, higher alignment accuracy is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 This is a flowchart of a complex routing preprocessing method based on 3D point cloud measurement and registration technology for this application;

[0047] Figure 2 This is a structural schematic diagram of a complex routing preprocessing system based on 3D point cloud measurement and registration technology for this application;

[0048] Figure 3 This is a structural schematic diagram of an electronic device of the present application.

[0049] icon:

[0050] 1. Acquisition module; 2. Cropping module; 3. Downsampling module; 4. Normal estimation module; 5. Feature extraction module; 6. Coarse registration module; 7. Fine registration module; 8. Routing design module; 9. Processor; 10. Memory; 11. Communication interface. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0052] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0053] Example

[0054] The existing technology uses tape measures or infrared distance detectors to measure the required wiring dimensions of steel pipes, but in the case of cross-sectional, dense and narrow areas, they are usually blocked and cannot be measured. In addition, the existing technology is also difficult to measure when there are long distances and multi-angle wiring bends at the target wiring location, making it impossible to perform pipeline wiring settings.

[0055] In view of this, an embodiment of the present application provides a complex routing preprocessing method and system based on 3D point cloud measurement and registration technology, which can generate point cloud images in complex environments to obtain wiring dimensions to facilitate users to design pipeline routing.

[0056] Please refer to Figure 1 In a first aspect, the present application provides a complex routing preprocessing method based on 3D point cloud measurement and registration technology, comprising:

[0057] S1: Use a binocular depth camera to shoot the target routing position in segments, and each shot generates a segment of point cloud data to obtain multiple segments of point cloud data;

[0058] Specifically, the binocular depth camera captures the target route position through two lenses to generate a point cloud image. The point cloud in the point cloud image is a collection of three-dimensional coordinates of a series of points in space, and these points represent the surface of objects in the scene. When the area of ​​the target route position is large, or the distance is too long, a single shot may not be able to fully or accurately obtain all the point cloud data of the target object while ensuring clarity. In order to solve this problem, the present application adopts a segmented shooting strategy, that is, shooting the target multiple times at different positions or angles, and each shot obtains a part of the point cloud data of the target, thereby obtaining multiple segments of point cloud data to ensure that sufficient and high-quality point cloud data can be obtained for subsequent alignment.

[0059] S2: Crop repeated point cloud data in multiple point cloud data segments;

[0060] Specifically, before performing point cloud stitching (registration), it is sometimes necessary to first crop the same parts from multiple point cloud data. These same parts are usually the overlapping areas of two point clouds, which contain common features that can be used to align the two point clouds. The purpose of cropping is to reduce the amount of calculation of the subsequent stitching algorithm and improve the accuracy of stitching. By focusing only on the overlapping areas, it is easier to find the correspondence between the two point clouds. The repeated point cloud data can be cropped using an automatic cropping algorithm or manually. Considering that the automatic cropping algorithm may not be able to accurately identify the overlapping areas in all cases, this embodiment allows the user to use the selection box or selection tool in the point cloud processing software to manually select or mark the same parts of the two point clouds for cropping.

[0061] S3: Perform voxel grid filtering on the cropped point cloud data;

[0062] Specifically, through voxel grid filtering, the point cloud data is divided into small three-dimensional cubes (voxels), and then a representative point (usually the center point of the voxel, the centroid or a point in the point cloud) is selected in each voxel to represent the points in the entire voxel. In this way, the number of points in the point cloud can be significantly reduced, thereby reducing the amount of calculation for subsequent processing and speeding up the running speed. First, the key function setLeafSize (0.001, 0.001, 0.001) in the voxel grid filtering is used to smooth or lose all details or features smaller than 1 mm. Then, the setLeafSize function is used to set the size of the voxel, that is, the side length of each three-dimensional cube. In this embodiment, the size of the voxel is set to 0.001 meters (or 1 mm), and the point cloud data is divided into small cubes with a side length of 1 mm, and only one representative point is retained in each cube (voxel). The density of the point cloud is successfully reduced, thereby reducing the amount of calculation and speeding up the running speed.

[0063] S4: performing normal estimation on the point cloud data after voxel grid filtering to obtain the surface normal feature of each point in the point cloud;

[0064] Specifically, it is preferred to use the key function setKSearch(20) to set the number of nearest neighbor points considered when calculating the normal to 20, and each point estimates its normal based on the 20 nearest neighbor points around it. Through normal estimation, the normal vector of the surface is calculated for each point in the point cloud data. The normal vector is a vector perpendicular to the tangent plane of the point cloud surface at that point, which provides important information about the shape and direction of the point cloud surface. setKSearch(20) is a key parameter setting in normal estimation, which specifies the number of nearest neighbor points used when calculating the normal of each point. In this embodiment, K is set to 20, and the 20 nearest neighbor points around the point will be considered when calculating the normal of each point. These nearest neighbor points are used to fit a local plane or surface, and then the normal vector is calculated based on this local plane or surface. By setting the parameter setKSearch(20), the accuracy and stability of normal estimation are ensured.

[0065] S5: extract the local normal feature of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points;

[0066] Specifically, it is preferred to use the key function setKSearch(30), set the number of nearest neighbor points considered when calculating features to 30, and calculate the features of each point based on the 30 nearest neighbor points around it. setKSearch(30) is a key parameter setting in FPFH feature extraction, which specifies the number of nearest neighbor points used when calculating the FPFH features of each point. In this embodiment, K is set to 30, and the 30 nearest neighbor points around the point are considered when calculating the FPFH features of each point. These nearest neighbor points are used to define a local neighborhood, and the FPFH features are calculated based on this local neighborhood. By setting setKSearch(30), the detailed geometric structure of the point cloud can be captured while avoiding the introduction of irrelevant points.

[0067] S6: Based on the geometric relationship between each point and its neighboring points, the SAC-IA algorithm is used to calculate the initial transformation matrix between the point clouds. Based on the initial transformation matrix between the point clouds, the two complete point clouds are preliminarily registered using their overlapping point clouds, so that the two point clouds are preliminarily aligned globally.

[0068] Specifically, it is preferred to use the key function setMaxCorrespondenceDistance(0.01) to set the maximum distance threshold when determining corresponding point pairs. The distance between two points must be within 0.01 meters to be considered as valid corresponding points. The initial transformation matrix (including rotation and translation) between the two point clouds is calculated by the SAC-IA algorithm to align them. It should be understood that in the SAC-IA algorithm, the FPFH feature is used to identify the corresponding point pairs between the two point clouds. The FPFH feature is a local feature description method that can capture the local geometry and surface changes of the point cloud, so it helps to find matching point pairs between point clouds. By matching point pairs, the SAC-IA algorithm can estimate the initial transformation matrix required to transform the source point cloud to the target point cloud. This initial transformation matrix is ​​usually not very accurate, but it provides a good starting point for the subsequent fine alignment step. setMaxCorrespondenceDistance(0.01) is a parameter setting in the SAC-IA algorithm that specifies the maximum allowable distance between two points when determining corresponding point pairs. In this embodiment, the maximum corresponding point distance is set to 0.01 meters (i.e., 1 centimeter). This means that only when the distance between two points is less than or equal to 1 cm, they will be considered a matching point pair. By setting the setMaxCorrespondenceDistance(0.01) parameter, the accuracy and stability of the registration result can be ensured.

[0069] S7: Use the ICP point cloud registration algorithm to accurately register the two complete point clouds to obtain the final point cloud image of the target alignment position;

[0070] Specifically, the parameters of the ICP point cloud registration algorithm are preferably set as follows: setMaximumIterations(100), setTransformationEpsilo(1e-20), and setEuclideanFitnessEpsilon(1e-20). By setMaximumIterations(100), the maximum number of iterations of the ICP algorithm is set to 100. That is, the algorithm will try to iterate up to 100 times to find the best alignment. If the maximum number of iterations is reached and the convergence condition is still not met, the algorithm will stop. setTransformationEpsilon(1e-20) and setEuclideanFitnessEpsilon(1e-20) set two very small thresholds to determine whether the ICP algorithm has converged. setTransformationEpsilon controls the amount of change in the transformation matrix, while setEuclideanFitnessEpsilon controls the amount of error between point clouds. If the amount of change in these two values ​​in consecutive iterations is less than the specified threshold, the algorithm is considered to have converged and the iteration stops. These very small thresholds (1e-20) mean that the ICP algorithm has to find the alignment between the source and target point clouds very accurately. This allows for higher registration accuracy.

[0071] In summary, the present invention uses a binocular depth camera to obtain a point cloud image of the target routing position. When encountering an obstructed situation, the position and angle of the binocular depth camera can be changed to perform segmented shooting to obtain multiple point cloud data. The acquisition operation is convenient and does not require much physical work. It completely solves the routing obstacle problem, ignores obstacles and the narrowness of space, and has high measurement accuracy. For situations where the target routing position has long-distance and multi-angle routing bends, the present invention can also obtain multiple segments of point cloud data by performing segmented shooting of the target routing position, and then by cropping the repeated point cloud data, the calculation amount of the subsequent splicing algorithm is reduced, and the accuracy of the splicing is improved. By performing voxel grid filtering on the cropped point cloud data, the calculation amount is further reduced and the running speed is accelerated. By setting the parameter setKSearch(20) for normal estimation, the accuracy and stability of normal estimation are ensured. By setting setKSearch(30) for feature extraction, the detailed geometric structure of the point cloud can be captured while avoiding the introduction of irrelevant points. By setting the parameter setMaxCorrespondenceDistance(0.01) for coarse registration, the accuracy and stability of the registration result are ensured. By setting the parameters of the ICP point cloud registration algorithm for fine registration, higher registration accuracy is obtained.

[0072] Please refer to Figure 2In a second aspect, the present invention provides a complex routing preprocessing system based on 3D point cloud measurement and registration technology, comprising:

[0073] The acquisition module 1 is used to shoot the target routing position in segments through a binocular depth camera, and each shooting generates a segment of point cloud data to obtain multiple segments of point cloud data;

[0074] A cutting module 2 is used to cut repeated point cloud data from multiple point cloud data segments;

[0075] Downsampling module 3, used for performing voxel grid filtering on the cropped point cloud data;

[0076] Normal estimation module 4, used for performing normal estimation on the point cloud data after voxel grid filtering to obtain the surface normal feature of each point in the point cloud;

[0077] A feature extraction module 5 is used to extract the local normal feature of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points;

[0078] A coarse registration module 6 is used to calculate the initial transformation matrix between the point clouds based on the geometric relationship between each point and its neighboring points using the SAC-IA algorithm, and based on the initial transformation matrix between the point clouds, perform preliminary registration between two complete point clouds using their overlapping point clouds, so that the two point clouds are preliminarily aligned globally;

[0079] A precise registration module 7 is used to precisely register two complete point clouds using an ICP point cloud registration algorithm to obtain a final point cloud image of the target alignment position;

[0080] The routing design module 8 is used to output the wiring dimensions of the target routing position according to the final point cloud image for routing design.

[0081] Specifically, for the specific implementation of the complex routing preprocessing system based on 3D point cloud measurement and registration technology, please refer to the implementation of the complex routing preprocessing method based on 3D point cloud measurement and registration technology mentioned above, and no further details will be given here.

[0082] Please refer to Figure 3 In a third aspect, the present invention provides an electronic device, comprising:

[0083] A memory 10, used for storing one or more programs;

[0084] Processor 9; the processor 9 is connected to the memory 10 via a communication interface 11.

[0085] When the one or more programs are executed by the processor, all or part of the above methods are implemented.

[0086] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements all or part of the above-mentioned method when executed by a processor.

[0087] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A complex routing preprocessing method based on 3D point cloud measurement and registration technology, characterized in that: include: The target routing position is photographed in segments by using a binocular depth camera, and each shot generates a segment of point cloud data to obtain multiple segments of point cloud data; Crop repeated point cloud data in multiple point cloud data segments; Perform voxel grid filtering on the cropped point cloud data; The normal line of the point cloud data after voxel grid filtering is estimated to obtain the surface normal line feature of each point in the point cloud; Extract the local normal features of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points; Based on the geometric relationship between each point and its neighboring points, the SAC-IA algorithm is used to calculate the initial transformation matrix between the point clouds. Based on the initial transformation matrix between the point clouds, the overlapping point clouds of the two complete point clouds are used to perform preliminary registration, so that the two point clouds are initially aligned globally. Use the ICP point cloud registration algorithm to accurately register the two complete point clouds to obtain the final point cloud image of the target alignment position; The wiring dimensions of the target routing position are output according to the final point cloud image for routing design.

2. According to claim 1, a complex routing preprocessing method based on 3D point cloud measurement and registration technology is characterized in that: The steps of cutting out repeated point cloud data in multi-segment point cloud data include: Manually select the same point cloud data and crop it.

3. The complex routing preprocessing method based on 3D point cloud measurement and registration technology according to claim 2 is characterized in that: The steps of voxel grid filtering on the cropped point cloud data include: Use the key function setLeafSize(0.001,0.001,0.001) in the voxel grid filter to smooth or lose details or features smaller than 1 mm.

4. The complex routing preprocessing method based on 3D point cloud measurement and registration technology according to claim 1 is characterized in that: The steps of performing normal estimation on the point cloud data after voxel grid filtering to obtain the surface normal feature of each point in the point cloud include: The key function setKSearch(20) is used to set the number of nearest neighbor points considered when calculating the normal to 20. Each point estimates its normal based on the 20 nearest neighbor points around it.

5. The complex routing preprocessing method based on 3D point cloud measurement and registration technology according to claim 1 is characterized in that: The steps of extracting the local normal feature of each point and calculating the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points include: Use the key function setKSearch(30) to set the number of nearest neighbor points considered when calculating features to 30. Each point calculates its features based on the 30 nearest neighbor points around it.

6. The complex routing preprocessing method based on 3D point cloud measurement and registration technology according to claim 1 is characterized in that: The steps to calculate the initial transformation matrix between point clouds using the SAC-IA algorithm include: Use the key function setMaxCorrespondenceDistance(0.01) to set the maximum distance threshold when determining corresponding point pairs. The distance between two points must be within 0.01 meters to be considered a valid corresponding point.

7. The complex routing preprocessing method based on 3D point cloud measurement and registration technology according to claim 1 is characterized in that: The steps of using the ICP point cloud registration algorithm to accurately register two complete point clouds to obtain the final point cloud image of the target alignment position include: Set the parameters of the ICP point cloud registration algorithm to: setMaximumIterations(100), setTransformationEpsilo(1e-20), setEuclideanFitnessEpsilon(1e-20).

8. A complex routing preprocessing system based on 3D point cloud measurement and registration technology, characterized in that: include: The acquisition module is used to shoot the target routing position in segments through a binocular depth camera, and each shooting generates a segment of point cloud data to obtain multiple segments of point cloud data; A cropping module is used to crop repeated point cloud data in multiple point cloud data segments; Downsampling module, used to perform voxel grid filtering on the cropped point cloud data; The normal estimation module is used to estimate the normal of the point cloud data after voxel grid filtering to obtain the surface normal features of each point in the point cloud; A feature extraction module is used to extract the local normal feature of each point, and calculate the descriptor of the local geometric structure of the point cloud based on the surface normal of each point to obtain the geometric relationship between each point and its neighboring points; The coarse registration module is used to calculate the initial transformation matrix between point clouds based on the geometric relationship between each point and its neighboring points using the SAC-IA algorithm. Based on the initial transformation matrix between point clouds, the initial registration is performed between two complete point clouds using their overlapping point clouds, so that the two point clouds are initially aligned globally. The precise registration module is used to precisely register two complete point clouds using the ICP point cloud registration algorithm to obtain the final point cloud image of the target alignment position; The routing design module is used to output the wiring dimensions of the target routing position according to the final point cloud image for routing design.

9. An electronic device, characterized in that: include: A memory for storing one or more programs; processor; When the one or more programs are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images