A method and system for surface reconstruction of a slender curved and twisted member

Abstract

This invention discloses a method and system for surface reconstruction of elongated curved and torsional components, belonging to the field of surface reconstruction technology for curved and torsional components. The method includes: inputting point cloud data of the component to obtain an initial point cloud, and preprocessing the initial point cloud to obtain a processed point cloud; classifying the processed point cloud according to the geometric features of the component to obtain inner and outer surface point clouds and side surface point clouds; extracting data from the inner and outer surface point clouds to obtain outer surface point clouds and inner surface point clouds, wherein the outer surface point cloud, inner surface point cloud, and side surface point cloud constitute the target surface point cloud; and sequentially performing multiple principal component analysis, density partitioning, and spline interpolation fitting on the target surface point cloud to obtain the target surface contour boundary line and the target surface. This invention achieves the construction of the contour boundary line and curved surface of elongated components, improves the construction accuracy, and meets practical engineering needs.

Description

Technical Field

[0001] This invention relates to the field of surface reconstruction technology for bent and torsional components, and in particular to a method and system for surface reconstruction of narrow bent and torsional components. Background Technology

[0002] In the field of civil engineering, the design of building envelopes or decorative structures such as glass curtain walls requires the measurement and positioning of the main structural components. However, most public buildings have complex shapes, and traditional measurement methods cannot achieve the desired results. Therefore, laser scanning technology is needed to obtain point cloud data for reconstruction.

[0003] Traditional point cloud processing methods face numerous challenges when processing point cloud data of narrow, curved, and tortuous structural members, as follows:

[0004] (1) Because the cross-sectional dimensions of such components are small and the point cloud is relatively thin, after deducting the point cloud of the boundary contour line, there is a lack of effective surface information, making it difficult to fully present its geometric features. Traditional algorithms are not ideal for surface reconstruction.

[0005] (2) When the building is tall, the point cloud distribution on the top area of ​​the outer surface is relatively sparse due to the shooting angle, that is, the point cloud quality is poor, and some information on the inner surface is even missing. During conventional noise reduction processing, this can easily lead to the loss of key data, affecting the integrity and usability of the data;

[0006] (3) When the algorithm cannot meet the requirements, it is difficult to manually process the bending and twisting components. Determining their surface requires a large amount of point cloud data, and the workload of data processing is huge.

[0007] In summary, traditional point cloud processing algorithms are mostly designed for regular components and struggle to achieve high-precision reconstruction of the contours and surface details of elongated components. When faced with such complex components, the limitations of traditional processing methods become apparent, failing to meet actual engineering needs. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for surface reconstruction of narrow, curved and twisted components, which solves the problems of unsatisfactory surface reconstruction results, lack of data integrity and usability, and large workload of traditional algorithms. It realizes the construction of the outline boundary line and curved surface of narrow components, improves the construction accuracy, and meets the actual needs of engineering.

[0009] To achieve the above objectives, the present invention provides a method for surface reconstruction of a narrow, elongated, bent, or torsional member, comprising the following steps:

[0010] S1. Input the point cloud data of the component to obtain the initial point cloud of the component, and preprocess the initial point cloud of the component to obtain the processed point cloud of the component;

[0011] S2. Based on the geometric characteristics of the component, the point cloud of the component is processed to perform surface classification to obtain a surface point cloud; the surface point cloud includes: inner and outer surface point clouds and side surface point clouds;

[0012] S3. Extract data from the inner and outer surface point clouds to obtain the outer surface point cloud and the inner surface point cloud; the outer surface point cloud, the inner surface point cloud and the side surface point cloud constitute the target surface point cloud.

[0013] S4. Perform multiple principal component analysis, density division processing and spline interpolation fitting on the target surface point cloud in sequence to obtain the target surface contour boundary line and the target surface; the target surface contour boundary line includes: outer surface contour boundary line, inner surface contour boundary line and side surface contour boundary line; the target surface includes: outer surface, inner surface and side surface.

[0014] Preferably, the specific content of the component processed point cloud obtained by preprocessing the initial point cloud of the component in S1 includes:

[0015] The point cloud normal vector is obtained by calculating the normal vector of the point cloud data using the Python-Open3D library; the point cloud normal vector points to infinity.

[0016] The component processing point cloud includes: the component initial point cloud and the point cloud normal vector.

[0017] Preferably, in step S2, the surface of the component point cloud is classified according to the geometric features of the component, and the specific content of the surface point cloud includes:

[0018] Based on the geometric characteristics of the component, determine the threshold angle between the point cloud normal vector and the horizontal plane;

[0019] Based on the angle threshold between the point cloud normal vector and the horizontal plane, the surface point cloud of the component is classified to obtain the surface point cloud.

[0020] Preferably, in step S3, data extraction is performed on the point clouds of the inner and outer surfaces to obtain the specific contents of the point clouds of the outer and inner surfaces, including:

[0021] Principal component analysis was performed on the point clouds of the inner and outer surfaces to determine the direction of the first principal component.

[0022] Construct a projection plane with the direction of the first principal component as the horizontal axis and the z-axis as the vertical axis;

[0023] Projecting the point clouds of the inner and outer surfaces onto the projection plane yields two-dimensional point clouds of the inner and outer surfaces;

[0024] Principal component analysis was performed on the two-dimensional point clouds of the inner and outer surfaces to determine the s-axis along the length direction and the t-axis along the height direction of the two-dimensional point clouds of the inner and outer surfaces.

[0025] Determine the orientation of the height direction axis t, as well as the maximum and minimum coordinate values ​​of the two-dimensional point clouds on the inner and outer surfaces along the height direction axis, and define a distance threshold.

[0026] Based on the orientation of the height direction axis t-axis, the maximum and minimum coordinate values, and the distance threshold, the two-dimensional point clouds of the inner and outer surfaces are filtered to obtain the outer surface point cloud and the inner and outer surface point clouds.

[0027] Preferably, the two-dimensional point clouds of the inner and outer surfaces are filtered based on the orientation of the height direction axis t, the maximum and minimum coordinate values, and the distance threshold, resulting in the outer surface point cloud and the specific contents of the inner and outer surface point clouds, including:

[0028] If the height direction axis t-axis points upward, then the coordinate value of the height direction axis t-axis is determined. The point cloud with a coordinate value greater than the difference between the maximum value and the distance threshold is the outer surface point cloud.

[0029] If the height direction axis t-axis points upward, then the point cloud with a height direction axis t-axis coordinate value less than the sum of the minimum coordinate value and the distance threshold is the inner surface point cloud.

[0030] If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. The point cloud with a coordinate value greater than the difference between the maximum value and the distance threshold is the inner surface point cloud.

[0031] If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. Point clouds with coordinate values ​​less than the sum of the minimum coordinate value and the distance threshold are the outer surface point clouds.

[0032] Preferably, in S4, multiple principal component analysis, density partitioning, and spline interpolation fitting are performed sequentially on the target surface point cloud to obtain the target surface contour boundary line and the specific content of the target surface, including:

[0033] Principal component analysis is performed on the point cloud of the target surface to obtain the axial vector;

[0034] The target surface point cloud is divided into n equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n coordinate systems, i.e., coordinate system U. n V n W n The coordinate system U n V n W n Middle U n V is the length axis. n W is the width axis. n The normal axis;

[0035] Take U n V is the horizontal axis. n Construct a projection plane of the point cloud into n equal parts, with the vertical axis as the ordinate;

[0036] Project each equal part of the point cloud onto the n equal parts of the point cloud projection plane, and along U n Along the horizontal axis, each equal portion of the projected point cloud is divided into m slices; n and m are both positive integers, and n is less than m;

[0037] Determine the boundary points in each slice; the point cloud coordinates of the boundary points along the vertical axis are the maximum and minimum values, and the point cloud coordinates along the normal axis are the maximum values.

[0038] Construct the first and second contour lines based on the boundary points;

[0039] By transforming the coordinates, the boundary points on the first and second contour lines are transformed to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour.

[0040] Spline interpolation fitting is performed on the boundary lines of each equal part of the point cloud contour to obtain the boundary lines of the target surface contour;

[0041] The surface formed by the boundary lines of the target surface contour is the target surface.

[0042] Preferably, the specific details of constructing the first and second contour lines based on the boundary points include:

[0043] Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line;

[0044] Connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum and the point cloud coordinates in the normal direction are at their maximum to obtain the second contour line.

[0045] The present invention also provides a surface reconstruction system for a narrow, elongated, bent, or torsional member, comprising:

[0046] The processing module is used to input the point cloud data of the component to obtain the initial point cloud of the component, and to preprocess the initial point cloud of the component to obtain the processed point cloud of the component.

[0047] The classification module is used to classify the surface of the component processing point cloud according to the geometric features of the component to obtain a surface point cloud; the surface point cloud includes: inner and outer surface point clouds and side surface point clouds;

[0048] The extraction module is used to extract data from the inner and outer surface point clouds to obtain the outer surface point cloud and the inner surface point cloud; the outer surface point cloud, the inner surface point cloud and the side surface point cloud constitute the target surface point cloud.

[0049] The construction module is used to sequentially perform multiple principal component analysis, density division processing, and spline interpolation fitting on the point cloud of the target surface to obtain the target surface contour boundary line and the target surface; the target surface contour boundary line includes: outer surface contour boundary line, inner surface contour boundary line, and side surface contour boundary line; the target surface includes: outer surface, inner surface, and side surface.

[0050] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor, when calling the computer program in the memory, implements the content of the above-described method for reconstructing the surface of a narrow, curved, and twisted component.

[0051] The present invention also provides a storage medium storing computer-executable instructions, which, when loaded and executed by a processor, implement the above-described method for surface reconstruction of a narrow, curved, and twisted component.

[0052] In summary, the surface reconstruction method and system for narrow, curved, and twisted components of the present invention, compared with traditional technologies, has the following advantages: By sequentially preprocessing, classifying, extracting data, performing principal component analysis, density partitioning, and spline interpolation fitting on the point cloud data of the input component, the target surface contour boundary line and the target surface are obtained. This solves the problems of unsatisfactory surface reconstruction processing effect, lack of data integrity and usability, and large workload of traditional algorithms. It achieves high-precision reconstruction of the contour boundary line and curved surface of narrow, curved components, improves construction accuracy, simplifies the construction process, and can meet the actual needs of engineering.

[0053] The technical method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0054] Figure 1 This is a flowchart of a method for surface reconstruction of a narrow, tortuous component according to the present invention;

[0055] Figure 2 This is a 3D isometric view of the initial point cloud of the component in this invention;

[0056] Figure 3 This is a three-dimensional isometric view of the outer surface contour line in this invention;

[0057] Figure 4 This is a three-dimensional isometric view of the outer surface in this invention;

[0058] Figure 5 This is a block diagram of a surface reconstruction system for a narrow, tortuous component according to the present invention. Detailed Implementation

[0059] The technical method of the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0060] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0061] Techniques, systems, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the instruction manual.

[0062] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0063] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0064] like Figure 1 As shown, this invention provides a method for surface reconstruction of a slender, torsional member, which is based on the geometric features of the slender member and point cloud multi-principal component analysis, including:

[0065] Step S1: Input the point cloud data of the component to obtain the initial point cloud of the component, such as... Figure 2 As shown, the component's initial point cloud is preprocessed to obtain the component's processed point cloud.

[0066] Furthermore, the specific content of the component processing point cloud preprocessing in step S1 includes: calculating the normal vector of the point cloud data according to the Python-Open3D library to obtain the point cloud normal vector. The point cloud normal vector points to infinity; the component processing point cloud includes: the component initial point cloud and the point cloud normal vector.

[0067] Step S2: Based on the geometric characteristics of the component, perform surface classification on the component's point cloud to obtain a surface point cloud. The surface point cloud includes: inner and outer surface point clouds and side surface point clouds.

[0068] Furthermore, in step S2, the surface of the component point cloud is classified according to the geometric features of the component, and the specific content of the surface point cloud includes:

[0069] Based on the geometric characteristics of the component, determine the threshold angle between the point cloud normal vector and the horizontal plane.

[0070] Based on the angle threshold between the point cloud normal vector and the horizontal plane, surface classification is performed on the component's processed point cloud to obtain the surface point cloud. Specifically, the angle threshold between the side surface point cloud normal vector and the horizontal plane is... θ The threshold value is 0; while the angle threshold between the normal vectors of the inner and outer surface point clouds and the horizontal plane is 0. θ The included angle threshold is determined by the geometric characteristics of the component. θ The size cannot be too small to avoid unclear classification and confusion between different surfaces; the included angle threshold... θ It shouldn't be too large either, to avoid misjudgment.

[0071] Step S3: Extract data from the inner and outer surface point clouds to obtain the outer surface point cloud and the inner surface point cloud. The outer surface point cloud, the inner surface point cloud, and the side surface point cloud constitute the target surface point cloud.

[0072] Furthermore, step S3 can be replaced by the following steps S301~S306:

[0073] Step S301: Perform principal component analysis on the point clouds of the inner and outer surfaces to determine the direction of the first principal component. The principal component analysis used is conventional principal component analysis (PCA), a traditional and mature mathematical processing method that can stably extract the main distribution direction of the point cloud.

[0074] Step S302: Construct a projection plane with the direction of the first principal component as the horizontal axis and the z-axis as the vertical axis.

[0075] Step S303: Project the inner and outer surface point clouds onto the projection plane to obtain the two-dimensional point clouds of the inner and outer surfaces.

[0076] Step S304: Perform principal component analysis on the two-dimensional point clouds of the inner and outer surfaces to determine the s-axis along the length direction and the t-axis along the height direction of the two-dimensional point clouds of the inner and outer surfaces.

[0077] Step S305: Determine the orientation of the height direction axis t, and the maximum and minimum coordinate values ​​of the two-dimensional point clouds on the inner and outer surfaces along the height direction axis, while defining a distance threshold. δ The distance threshold among them. δ This is equivalent to a "discrimination bandwidth," which is mainly set based on the component's cross-sectional dimensions and point cloud density, along with a set distance threshold. δ The goal is to effectively filter out noise while maintaining the integrity of the outer surface data. Furthermore, when the point cloud density is high and the component size is small, the distance threshold... δ The value of is reduced accordingly; when the point cloud density is low and the component size is large, the distance threshold is increased. δ The value of should be increased appropriately.

[0078] Step S306: Based on the orientation of the height direction axis t-axis, the maximum and minimum coordinate values, and the distance threshold, filter the two-dimensional point clouds of the inner and outer surfaces to obtain the outer surface point cloud and the inner surface point cloud. Specific details include:

[0079] If the height direction axis t is pointing upwards, then the coordinate value of the height direction axis t is determined. Point clouds with coordinate values ​​greater than the difference between the maximum value and the distance threshold are considered as outer surface point clouds.

[0080] If the height direction axis t-axis points upward, then the coordinate value of the height direction axis t-axis is determined. Point clouds whose coordinate values ​​are less than the sum of the minimum coordinate value and the distance threshold are considered inner surface point clouds.

[0081] If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. Point clouds with coordinate values ​​greater than the difference between the maximum value and the distance threshold are considered inner surface point clouds.

[0082] If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. Point clouds with coordinate values ​​less than the sum of the minimum coordinate value and the distance threshold are the outer surface point clouds.

[0083] Step S4: Perform multiple principal component analysis, density partitioning, and spline interpolation fitting sequentially on the target surface point cloud to obtain the target surface contour boundary line and the target surface. The target surface contour boundary line includes: the outer surface contour boundary line, the inner surface contour boundary line, and the side surface contour boundary line; the target surface includes: the outer surface, the inner surface, and the side surface.

[0084] Furthermore, step S4 specifically includes the following steps:

[0085] Step S401: Perform principal component analysis on the point cloud of the target surface to obtain the axial vector.

[0086] Step S402: Divide the target surface point cloud into n equal parts along the axial vector direction, and perform principal component analysis on each part to obtain n coordinate systems, i.e., coordinate system U. n V n W n Among them, the coordinate system U n V n W n Middle U n V is the length axis. n W is the width axis. n The normal axis is denoted by . The directions of the first principal components of the n-part point cloud are all along the axis.

[0087] Step S403, with U n V is the horizontal axis. n Using the vertical axis, construct an n-part projection plane for the point cloud.

[0088] Step S404: Project each equal part of the point cloud onto the n equal parts of the point cloud projection plane, and along U n Along the horizontal axis, each equal portion of the projected point cloud is divided into m slices. Here, n and m are both positive integers, with n less than m. The values ​​of n and m depend on the curvature of the component and can be adjusted. When the curvature of the component is high, the values ​​of n and m are larger; when the curvature of the component is low, the values ​​of n and m are smaller. When the curvature of the component is 0, the value of n is 1.

[0089] Step S405: Determine the boundary points in each slice. The point cloud coordinates along the vertical axis of the boundary points are the maximum and minimum values, while the point cloud coordinates along the normal axis are the maximum values.

[0090] Step S406, constructing the first and second contour lines based on the boundary points, includes the following:

[0091] Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line.

[0092] Connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum and the point cloud coordinates in the normal direction are at their maximum to obtain the second contour line.

[0093] Step S407: Through coordinate transformation, the boundary points on the first and second contour lines are transformed to the global coordinate system to obtain the boundary lines of each equal part of the point cloud contour.

[0094] Step S408: Perform spline interpolation fitting on the boundary lines of each equal part of the point cloud to obtain the boundary lines of the target surface.

[0095] Step S409: The surface formed by the boundary lines of the target surface contour is the target surface.

[0096] An exemplary embodiment of the present invention provides specific steps for performing multiple principal component analysis, density partitioning, and spline interpolation fitting sequentially on the point cloud of the outer surface to obtain the outer surface contour boundary line and the outer surface, including:

[0097] Principal component analysis is performed on the point cloud of the outer surface to obtain the axial vector.

[0098] The point cloud on the outer surface is divided into n1 equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n1 coordinate systems, i.e., coordinate system U. n 1 V n 1 W n 1 Among them, the coordinate system U n 1 V n1 W n 1 Middle U n 1 V is the length axis. n 1 W is the width axis. n 1 Let n be the normal axis. Where n1 is 4.

[0099] Take U n 1 V is the horizontal axis. n 1 Using the vertical axis, construct an n1-part point cloud projection plane.

[0100] Project each equal part of the point cloud onto the n1 equal part point cloud projection plane, and along U n 1 Along the horizontal axis, each equal portion of the projected point cloud is divided into m1 slices, where m1 is 100. Both n1 and m are positive integers, and n1 is less than m1, indicating a high slice density.

[0101] Identify the boundary points in each slice. These boundary points form a boundary point set. The point cloud coordinates along the vertical axis of each boundary point represent its maximum and minimum values, while the point cloud coordinates along the normal axis represent their maximum values.

[0102] Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line; connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the second contour line.

[0103] Connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum and the point cloud coordinates in the normal direction are at their maximum to obtain the second contour line.

[0104] By transforming coordinates, the boundary points on the first and second contour lines are converted to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour.

[0105] By fitting points on each equal part of the contour boundary line of the outer surface, a spline curve is obtained. A smooth curve is obtained by spline interpolation. Then, a few points are sampled on the smooth curve, the coordinates of the points are calculated, and the contour line of the outer surface is drawn based on the coordinates of all sampled points.

[0106] The surface formed by the outer surface contour boundary lines is the reconstructed outer surface.

[0107] Another exemplary embodiment of the present invention provides specific steps for sequentially performing multiple principal component analysis, density partitioning, and spline interpolation fitting on the inner surface point cloud to obtain the inner surface contour boundary line and the inner surface, including:

[0108] Principal component analysis was performed on the point cloud of the inner surface to obtain the axial vector.

[0109] The inner surface point cloud is divided into n² equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n² coordinate systems, i.e., coordinate system U. n 2 V n 2 W n 2 Among them, the coordinate system U n 2 V n 2 W n 2 Middle U n 2 V is the length axis. n 2 W is the width axis. n 2 Let n be the normal axis. Where n2 is 4.

[0110] Take U n 2 V is the horizontal axis. n 2 Using the vertical axis, construct an n2-part point cloud projection plane.

[0111] Project each equal part of the point cloud onto the n2 equal part point cloud projection plane, and along U n 2 Along the horizontal axis, each equal portion of the projected point cloud is divided into m² slices, where m² is 100. Both n² and m² are positive integers, and n² is less than m², indicating a high slice density.

[0112] Identify the boundary points in each slice. These boundary points form a boundary point set. The point cloud coordinates along the vertical axis of each boundary point represent its maximum and minimum values, while the point cloud coordinates along the normal axis represent their maximum values.

[0113] Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line; connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the second contour line.

[0114] By transforming coordinates, the boundary points on the first and second contour lines are converted to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour.

[0115] By fitting points on each equal part of the contour boundary line of the inner surface, a spline curve is obtained. A smooth curve is obtained by spline interpolation. Then, a few points are sampled on the smooth curve, the coordinates of the points are calculated, and the contour line of the inner surface is drawn based on the coordinates of all sampled points.

[0116] The surface formed by the inner surface contour boundary line is the reconstructed inner surface.

[0117] Another exemplary embodiment of the present invention provides specific steps for performing multiple principal component analysis, density partitioning, and spline interpolation fitting sequentially on the side surface point cloud to obtain the side surface contour boundary line and the side surface, including:

[0118] Principal component analysis was performed on the point cloud of the side surface to obtain the axial vector.

[0119] The side surface point cloud is divided into n3 equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n3 coordinate systems, i.e., coordinate system U. n 3 V n 3 W n 3 Among them, the coordinate system U n 3 V n 3 W n 3 Middle U n 3 V is the length axis. n 3 W is the width axis. n 3 Let n be the normal axis. Where n3 is 4.

[0120] Take U n 3 V is the horizontal axis. n 3 Using the vertical axis, construct an n3-part point cloud projection plane.

[0121] Project each equal part of the point cloud onto the n3 equal part point cloud projection plane, and along U n 3 Along the horizontal axis, each equal portion of the projected point cloud is divided into m³ slices. Here, m³ is 100. Both n³ and m³ are positive integers, and n³ is less than m³, indicating a high slice density.

[0122] Identify the boundary points in each slice. These boundary points form a boundary point set. The point cloud coordinates along the vertical axis of each boundary point represent its maximum and minimum values, while the point cloud coordinates along the normal axis represent their maximum values.

[0123] Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line; connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the second contour line.

[0124] By transforming coordinates, the boundary points on the first and second contour lines are converted to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour.

[0125] By fitting points on each equal part of the contour boundary line of the side surface, a spline curve is obtained. A smooth curve is obtained by spline interpolation. Then, a few points are sampled on the smooth curve, the coordinates of the points are calculated, and the contour line of the side surface is drawn based on the coordinates of all sampled points.

[0126] The surface formed by the boundary lines of the side surface contour is the reconstructed side surface.

[0127] Taking the construction results of the outer surface contour and the outer surface as an example, the 3D isometric view of the constructed outer surface contour is as follows: Figure 3 As shown, the constructed three-dimensional isometric view of the outer surface is as follows: Figure 4 As shown.

[0128] In an exemplary embodiment of the present invention, based on the m slices divided in step S404, the cross-sectional dimensions of the component (i.e., the component width and the component height) can also be measured, as detailed below:

[0129] Based on m1 slices defined when constructing the outer surface or m2 slices defined when constructing the inner surface, principal component analysis (PCA) is performed on each slice. The first principal component direction is determined to be the width direction, and the maximum and minimum coordinate values ​​along this direction are obtained. To improve computational efficiency, a preset interval parameter p can be used during PCA, and PCA is performed every p slices. The range of values ​​for the interval parameter p depends on the number of measurements. or The range is an integer, and the standard requires that the number of measured parts be at least 3p less than m1 or m2.

[0130] The difference between the maximum and minimum coordinates in the width direction is determined as the dimension in the width direction, i.e., the width value.

[0131] The average value, or component width, is calculated based on the width of each slice.

[0132] Based on the m³ slices divided during the construction of the side surface, principal component analysis (PCA) is performed on each slice. The first principal component direction is determined to be the height direction, and the maximum and minimum coordinate values ​​along this direction are obtained. To improve computational efficiency during PCA, a preset interval parameter q can be used, and PCA is performed every q slices. The range of values ​​for the interval parameter q can be determined based on the number of measurements. The range is an integer, and the standard requires that the number of measured parts be at least 3q less than m3.

[0133] The difference between the maximum and minimum coordinate values ​​in the height direction is determined as the dimension in the height direction, i.e., the height value.

[0134] The average value, i.e., the component height, is calculated based on the height value of each slice.

[0135] The width and height of a component are its cross-sectional dimensions.

[0136] In an exemplary embodiment of the present invention, the narrow component has a width of only 40mm and a length of 2090mm. The point cloud used consists of 26655 points. The largest horizontal axis coordinate is 4.3345m, the vertical axis coordinate is 4.3465m, and the vertical axis coordinate is 3.7415m. The smallest horizontal axis coordinate is 4.2485m, the vertical axis coordinate is 3.5475m, and the vertical axis coordinate is 2.1135m.

[0137] The present invention also provides a surface reconstruction system for elongated bent and torsional components, such as Figure 5 As shown, it includes:

[0138] The processing module is used to input the point cloud data of the component to obtain the initial point cloud of the component, and to preprocess the initial point cloud of the component to obtain the processed point cloud of the component.

[0139] The classification module is used to classify the surface of the component's processed point cloud based on the component's geometric features, resulting in a surface point cloud. This surface point cloud includes: inner and outer surface point clouds and side surface point clouds.

[0140] The extraction module is used to extract data from the inner and outer surface point clouds, obtaining the outer surface point cloud and the inner surface point cloud. The outer surface point cloud, the inner surface point cloud, and the side surface point cloud constitute the target surface point cloud.

[0141] The module is used to sequentially perform multiple principal component analysis, density partitioning, and spline interpolation fitting on the target surface point cloud to obtain the target surface contour boundary line and the target surface. The target surface contour boundary line includes: outer surface contour boundary line, inner surface contour boundary line, and side surface contour boundary line; the target surface includes: outer surface, inner surface, and side surface.

[0142] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned method for reconstructing the surface of a narrow, curved, and twisted component when it calls the computer program in the memory.

[0143] The present invention also provides a storage medium storing computer-executable instructions, which, when loaded and executed by a processor, implement the above-mentioned method for surface reconstruction of a narrow, curved, and twisted component.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical methods of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical methods of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical methods to deviate from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for reconstructing the surface of a narrow, tortuous component, characterized in that, The method for reconstructing the surface of a narrow, tortuous component includes the following steps: S1. Input the point cloud data of the component to obtain the initial point cloud of the component, and preprocess the initial point cloud of the component to obtain the processed point cloud of the component; S2. Based on the geometric characteristics of the component, the point cloud of the component is processed to perform surface classification to obtain a surface point cloud; the surface point cloud includes: inner and outer surface point clouds and side surface point clouds; S3. Extract data from the inner and outer surface point clouds to obtain the outer surface point cloud and the inner surface point cloud; the outer surface point cloud, the inner surface point cloud and the side surface point cloud constitute the target surface point cloud. S4. Perform multiple principal component analysis, density division processing, and spline interpolation fitting sequentially on the target surface point cloud to obtain the target surface contour boundary line and the target surface; the target surface contour boundary line includes: outer surface contour boundary line, inner surface contour boundary line, and side surface contour boundary line; the target surface includes: outer surface, inner surface, and side surface; In S4, multiple principal component analysis, density partitioning, and spline interpolation fitting are performed sequentially on the target surface point cloud to obtain the target surface contour boundary line and the specific content of the target surface, including: Principal component analysis is performed on the point cloud of the target surface to obtain the axial vector; The target surface point cloud is divided into n equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n coordinate systems, i.e., coordinate system U. n V n W n The coordinate system U n V n W n Middle U n V is the length axis. n W is the width axis. n The normal axis; Take U n V is the horizontal axis. n Construct a projection plane of the point cloud into n equal parts, with the vertical axis as the ordinate; Project each equal part of the point cloud onto the n equal parts of the point cloud projection plane, and along U n Along the horizontal axis, each equal portion of the projected point cloud is divided into m slices; n and m are both positive integers, and n is less than m; Determine the boundary points in each slice; the point cloud coordinates of the boundary points along the vertical axis are the maximum and minimum values, and the point cloud coordinates along the normal axis are the maximum values. Construct the first and second contour lines based on the boundary points; By transforming the coordinates, the boundary points on the first and second contour lines are transformed to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour. Spline interpolation fitting is performed on the boundary lines of each equal part of the point cloud contour to obtain the boundary lines of the target surface contour; The surface formed by the boundary lines of the target surface contour is the target surface.

2. The method for surface reconstruction of a narrow, elongated, bent, and twisted component according to claim 1, characterized in that, The specific content of the component processed point cloud obtained by preprocessing the initial point cloud of the component in S1 includes: The point cloud normal vector is obtained by calculating the normal vector of the point cloud data using the Python-Open3D library; the point cloud normal vector points to infinity. The component processing point cloud includes: the component initial point cloud and the point cloud normal vector.

3. The method for surface reconstruction of a narrow, elongated, bent, and twisted component according to claim 2, characterized in that, In S2, the surface of the component point cloud is classified according to the component's geometric features, and the specific content of the surface point cloud includes: Based on the geometric characteristics of the component, determine the threshold angle between the point cloud normal vector and the horizontal plane; Based on the angle threshold between the point cloud normal vector and the horizontal plane, the surface point cloud of the component is classified to obtain the surface point cloud.

4. The method for surface reconstruction of a narrow, elongated, bent, and twisted component according to claim 1, characterized in that, In S3, data extraction is performed on the point clouds of the inner and outer surfaces, and the specific contents of the outer surface point cloud and the inner and outer surface point clouds are as follows: Principal component analysis was performed on the point clouds of the inner and outer surfaces to determine the direction of the first principal component. Construct a projection plane with the direction of the first principal component as the horizontal axis and the z-axis as the vertical axis; Projecting the point clouds of the inner and outer surfaces onto the projection plane yields two-dimensional point clouds of the inner and outer surfaces; Principal component analysis was performed on the two-dimensional point clouds of the inner and outer surfaces to determine the s-axis along the length direction and the t-axis along the height direction of the two-dimensional point clouds of the inner and outer surfaces. Determine the orientation of the height direction axis t, as well as the maximum and minimum coordinate values ​​of the two-dimensional point clouds on the inner and outer surfaces along the height direction axis, and define a distance threshold. Based on the orientation of the height direction axis t-axis, the maximum and minimum coordinate values, and the distance threshold, the two-dimensional point clouds of the inner and outer surfaces are filtered to obtain the outer surface point cloud and the inner and outer surface point clouds.

5. The method for surface reconstruction of a narrow, elongated, bent, and twisted component according to claim 4, characterized in that, Based on the orientation of the height direction axis (t-axis), the maximum and minimum coordinate values, and the distance threshold, the two-dimensional point clouds of the inner and outer surfaces are filtered to obtain the specific contents of the outer and inner surface point clouds, including: If the height direction axis t-axis points upward, then the coordinate value of the height direction axis t-axis is determined. The point cloud with a coordinate value greater than the difference between the maximum value and the distance threshold is the outer surface point cloud. If the height direction axis t-axis points upward, then the point cloud with a height direction axis t-axis coordinate value less than the sum of the minimum coordinate value and the distance threshold is the inner surface point cloud. If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. The point cloud with a coordinate value greater than the difference between the maximum value and the distance threshold is the inner surface point cloud. If the height direction axis t is pointing downwards, then the coordinate value of the height direction axis t is determined. Point clouds with coordinate values ​​less than the sum of the minimum coordinate value and the distance threshold are the outer surface point clouds.

6. The method for surface reconstruction of a narrow, elongated, bent, and twisted component according to claim 1, characterized in that, The specific steps for constructing the first and second contour lines based on the boundary points include: Connect the boundary points where the point cloud coordinates in the vertical direction are at their maximum values ​​and the point cloud coordinates in the normal direction are at their maximum values ​​to obtain the first contour line; Connect the boundary points where the point cloud coordinates in the vertical direction are at their minimum and the point cloud coordinates in the normal direction are at their maximum to obtain the second contour line.

7. A surface reconstruction system for a narrow, elongated, bent, and tortuous component, characterized in that, include: The processing module is used to input the point cloud data of the component to obtain the initial point cloud of the component, and to preprocess the initial point cloud of the component to obtain the processed point cloud of the component. The classification module is used to classify the surface of the component point cloud based on the component's geometric features, and obtain the surface point cloud. The surface point cloud includes: inner and outer surface point clouds and side surface point clouds; The extraction module is used to extract data from the inner and outer surface point clouds to obtain the outer surface point cloud and the inner surface point cloud; the outer surface point cloud, the inner surface point cloud and the side surface point cloud constitute the target surface point cloud. The construction module is used to sequentially perform multiple principal component analysis, density partitioning, and spline interpolation fitting on the target surface point cloud to obtain the target surface contour boundary line and the target surface. The target surface contour boundary line includes: outer surface contour boundary line, inner surface contour boundary line, and side surface contour boundary line; the target surface includes: outer surface, inner surface, and side surface. The specific content of obtaining the target surface contour boundary line and the target surface by sequentially performing multiple principal component analysis, density partitioning, and spline interpolation fitting on the target surface point cloud includes: Principal component analysis is performed on the point cloud of the target surface to obtain the axial vector; The target surface point cloud is divided into n equal parts along the axial vector direction, and principal component analysis is performed on each part to obtain n coordinate systems, i.e., coordinate system U. n V n W n The coordinate system U n V n W n Middle U n V is the length axis. n W is the width axis. n The normal axis; Take U n V is the horizontal axis. n Construct a projection plane of the point cloud into n equal parts, with the vertical axis as the ordinate; Project each equal part of the point cloud onto the n equal parts of the point cloud projection plane, and along U n Along the horizontal axis, each equal portion of the projected point cloud is divided into m slices; n and m are both positive integers, and n is less than m; Determine the boundary points in each slice; the point cloud coordinates of the boundary points along the vertical axis are the maximum and minimum values, and the point cloud coordinates along the normal axis are the maximum values. Construct the first and second contour lines based on the boundary points; By transforming the coordinates, the boundary points on the first and second contour lines are transformed to the global coordinate system, thus obtaining the boundary lines of each equal part of the point cloud contour. Spline interpolation fitting is performed on the boundary lines of each equal part of the point cloud contour to obtain the boundary lines of the target surface contour; The surface formed by the boundary lines of the target surface contour is the target surface.

8. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor, when calling the computer program in the memory, implements the content of the surface reconstruction method for a narrow, tortuous member as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores computer-executable instructions, which, when loaded and executed by a processor, implement the content of the surface reconstruction method for a narrow, curved, and twisted component as described in any one of claims 1 to 6.

Images