Method, system and medium for quickly extracting axis in three-dimensional laser point cloud of initial support of highway tunnel

Abstract

The application relates to a kind of highway tunnel primary support three-dimensional laser point cloud axis's fast extraction method, the method comprises the following steps: obtaining the three-dimensional laser point cloud of highway tunnel primary support, point cloud denoising processing is carried out, and the tunnel primary support point cloud block after denoising is extracted;The tunnel primary support point cloud block after denoising is projected on XOY and XOZ plane, forms XOY plane point cloud and XOZ plane point cloud, and carries out edge detection, extracts XOY plane boundary point cloud and XOZ plane boundary point cloud;Based on the XOY plane boundary point cloud and XOZ plane boundary point cloud, the middle axis point cloud coordinates on XOY plane and XOZ plane are calculated respectively using slice method;Based on the middle axis point cloud coordinates on XOY plane and XOZ plane, based on the relationship of adjacent point, determine the final three-dimensional middle axis point cloud coordinates.Compared with prior art, the application has the advantages of improving the accuracy of middle axis extraction.

Description

Technical Field

[0001] This invention relates to the field of point cloud processing for highway tunnels, and in particular to a rapid method for extracting the central axis of a three-dimensional laser point cloud of the initial support of a highway tunnel. Background Technology

[0002] In recent years, with the continuous advancement of highway tunnel construction, the geological conditions of tunnel sites have become increasingly complex, leading to a surge in geological disasters during construction, particularly landslides. Landslides can be broadly categorized into two types based on their mechanisms: landslides caused by unfavorable structural plane combinations and landslides caused by the plastic deformation of weak, fractured rock masses. For landslides caused by the plastic deformation of soft, fractured rock masses, since there is usually a relatively long deformation process before the landslide, rapid monitoring of the development of this deformation can provide early warning of tunnel landslides. Three-dimensional laser scanners, as a method that can accurately measure tunnel deformation, are increasingly being used in field applications.

[0003] To improve the efficiency of 3D laser scanning, a targetless 3D laser point cloud acquisition method has been proposed. By fitting the central axis of the tunnel point cloud data, the two sets of point cloud data are translated to the same coordinate system for deformation calculation. Among these methods, the key to accurately calculating tunnel deformation lies in how to quickly and accurately extract the tunnel's central axis.

[0004] Currently, there are several methods for extracting the centerline of a tunnel: Method 1: The tunnel is sliced ​​axially, and its outer contour is fitted with segmented circular arcs according to curvature. The center of each arc is determined, and finally, the axis is fitted using the least squares method to obtain the central axis (as disclosed in patent CN118096737B). This method requires initial determination of the tunnel axis and some manual processing using software such as ArcGIS and CloudCompare.

[0005] Method 2: The point cloud data is projected onto a plane (XOY direction), and the outline boundary of the two-dimensional planar tunnel point cloud is extracted. The axial and radial point cloud boundaries are separated by judging the long and short sides, and the plane centerline is further calculated. Finally, the z-direction coordinate is determined (as disclosed in patent application CN116934961A).

[0006] Method 3: The point cloud data is subjected to bidirectional planar projection, such as the XOY and YOZ directions. The point clouds in the two directions are segmented and processed to obtain the feature point clouds or central axis point clouds in the two directions respectively. Then, the data is fitted to obtain the three-dimensional central axis of the tunnel point cloud (such as the technology disclosed in patent applications CN116304764A and CN115035026A).

[0007] It is evident that current methods for extracting the centerline of 3D laser scanning point clouds of highway tunnels have the following main drawbacks: (1) Method 1: In the key steps of three-dimensional laser scanning point cloud processing, manual processing is still required, which is inefficient, difficult to operate, and cannot achieve rapid and automated extraction of the central axis point cloud.

[0008] (2) Method 2: There is still uncertainty in the extraction of the three-dimensional centerline coordinates by single-plane projection. Since the centerline coordinates obtained by single-plane projection only contain information in two dimensions, it is still difficult to obtain the third coordinate in the three-dimensional case. Generally, if the selected point cloud range is not large, it can be approximated that the tunnel slope is constant. In this case, the third coordinate of the point cloud is located in the same plane, and the calculation is relatively easy. However, this approach is prone to causing significant errors in the third coordinate when the distribution range of the point cloud changes greatly, which is not conducive to the accurate acquisition of the tunnel centerline.

[0009] (3) Method 3: Multiple fitting of point cloud will lead to a decrease in the accuracy of the central axis. Summary of the Invention

[0010] The purpose of this invention is to provide a rapid extraction method for the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel.

[0011] The objective of this invention can be achieved through the following technical solutions: A rapid method for extracting the centerline of the initial support of a highway tunnel using a 3D laser point cloud includes the following steps: The three-dimensional laser point cloud of the initial support of the highway tunnel is obtained, the point cloud is denoised, and the denoised tunnel initial support point cloud block is extracted. The denoised tunnel initial support point cloud block is projected onto the XOY and XOZ planes to form XOY plane point cloud and XOZ plane point cloud, and edge detection is performed to extract XOY plane boundary point cloud and XOZ plane boundary point cloud. Based on the boundary point clouds of the XOY plane and the XOZ plane, the coordinates of the central axis point cloud on the XOY plane and the XOZ plane are calculated using the slicing method. Based on the point cloud coordinates of the central axis on the XOY and XOZ planes, the final three-dimensional point cloud coordinates of the central axis are determined based on the nearest neighbor relationship.

[0012] Furthermore, the step of extracting the denoised tunnel initial pivot cloud blocks includes: Based on the three-dimensional laser point cloud of the initial support of the highway tunnel, the Euclidean distance growing algorithm is used for preliminary denoising to obtain the three-dimensional laser point cloud after preliminary denoising. Based on the 3D laser point cloud after initial denoising, the PCA algorithm is used for further denoising, and the Euclidean clustering algorithm is combined to extract the denoised tunnel initial support point cloud blocks.

[0013] Furthermore, the step of obtaining the preliminarily denoised 3D laser point cloud includes: 1) Mark all points in the three-dimensional laser point cloud as unclassified, denoted as point set α; 2) Take any point from the point set α. Place them into the point set Ω and mark them as classified; 3) Calculation points Are there any neighboring points within the growth distance r? If so, then move the point... If the points are placed into the set γ to be classified, proceed to step 4; otherwise, proceed to step 5. 4) Randomly select a point from the set of points to be classified, γ. Place it into the point set Ω, mark it as classified, and repeat step 3); 5) Take the points in the point set Ω as a cluster, and determine whether the number of points in the cluster is greater than the minimum size S of the cluster. If yes, keep the cluster; otherwise, discard the cluster. The minimum size S of the cluster is taken as 1 / k of the total number of point clouds n, where k is a set constant. 6) Return to step 3) until there are no more unclassified points, and output the cluster with the most points as the 3D laser point cloud after preliminary denoising.

[0014] Furthermore, the step of extracting the denoised tunnel initial pivot cloud blocks includes: For the preliminarily denoised 3D laser point cloud, the PCA algorithm is used to calculate the normal vector of each point cloud. ; Based on the normal vectors of each point cloud, the sign of the dot product is used to determine the ambiguity of the normal vectors, so as to unify the direction of all normal vectors and obtain point cloud normal vectors with unified direction. The expression for unifying the direction is as follows: , In the formula, For the first i The normal vector of a point The reference direction vector; Based on the point cloud normal vectors with unified orientation, any point is used as the query point. Calculate the query point With each point in the neighborhood set The included angle And sum up all the included angles to get the total included angle. ; Based on the sum of the included angles Calculate the average included angle And determine whether it satisfies If so, the query point If a point is considered noise, it is removed; otherwise, it is not considered noise and is not removed. This completes the denoising process, resulting in a denoised 3D laser point cloud. The set included angle threshold; For the denoised 3D laser point cloud, Euclidean clustering is used for block clustering, and the point cloud block with the largest number of points is taken as the denoised tunnel initial support point cloud block.

[0015] Furthermore, the Algha-Shape algorithm is used for edge detection.

[0016] Furthermore, the steps for obtaining the XOY plane boundary point cloud and the XOZ plane boundary point cloud include: (1) Take any point in the XOY plane point cloud Centered on, based on the rolling radius α The circle defined by 2 α Distance threshold, search all [distance thresholds] p The distance between the points does not exceed 2 α Let the points be denoted as point set Q; (2) Take any point in the point set Q According to the point and points coordinates and rolling radius Calculate For strings, Coordinates of the centers of two circles with radius and The expression for calculating the center coordinates is: , , , , In the formula, The radius of the rolling circle, ; (3) In the point set Q, calculate except p 1 Coordinates of all points other than the center point to the center of the circle o 1 distance L 1 Sum to the center coordinates o 2 distance L 2 ; (4) Determine when excluding p 1 Distance to all points other than point 1L 1 and distance L 2 Are they all greater than α ,if, p If a point is selected as the boundary point, then another point is selected from the point set Q. p 1 Return to step (2) to calculate the new distance. L 1 and distance L 2 And re-evaluate; after traversing all points in the point set Q, none of them satisfy the condition greater than 1. α The conditions for judgment p The point is not a boundary point; (5) Denote all boundary points as the boundary point cloud of the XOY plane; For the XOZ plane point cloud, repeat all the above steps to obtain all edge points on the XOZ plane, denoted as the XOZ plane edge point cloud.

[0017] Furthermore, the calculation steps for the coordinates of the central axis point cloud on the XOY and XOZ planes include: For the XOY plane boundary point cloud P, calculate its principal direction and center point C to obtain the principal direction vector. and slice direction vector ; Points on the XOY plane boundary point cloud p i To the main direction vector Projecting upwards, we obtain the representation points. p i Scalar of position on the principal axis s i The scalar s i The calculation expression is: , Set slice thickness d and slice step size ; scalar of minimum value s i As the starting position, at the slice step size Find the set slice center position k To obtain the points that satisfy the constraints. p i The set of point clouds representing tunnel sections, wherein the constraints are: , The points in the tunnel section point cloud set pi Projected onto the principal direction vector Above, we obtain a set of scalars. t i Sequence, and take the said t i scalars of sites in the sequence t i ’ As the centerline point corresponding to the slice; Delete the processed point cloud and continue to process the remaining point cloud according to the slicing method until the point cloud processing is completed. Finally, take the coordinates of all the central axis points in the XY coordinate system processed by the slicing method as the coordinates of the central axis point cloud on the XOY plane. For the boundary point cloud of the XOZ plane, repeat all the above steps to obtain the coordinates of the central axis point cloud on the XOZ plane.

[0018] Furthermore, the step of determining the final three-dimensional centerline point cloud coordinates includes: Define the coordinates of the centerline point cloud on the XOY plane as L1 and the coordinates of the centerline point cloud on the XOZ plane as L2; Find the coordinates of all points in L1. x ci ,y ci Converting ) to the centerline coordinates L3 of a 3D point cloud, denoted as ( x ci ,y ci , 0); Find any point in L2 that is in L1 y ci By taking the values ​​of n adjacent points, obtain the corresponding points in the point cloud coordinates of the central axis on the XOZ plane. z ci Find the value and find the nth one. z ci The average value is used as the coordinate L3 of the central axis of the three-dimensional point cloud. z The value confirms the coordinates of the central axis of a 3D point cloud. For the remaining points in L1, repeat all the above steps to confirm the coordinates of the central axis of all three-dimensional point clouds in L3, obtain the final coordinates of the three-dimensional central axis point cloud, and fit a straight line to obtain the central axis of the tunnel's initial support.

[0019] The present invention also provides a rapid extraction system for the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel, including a memory, a processor, and a program stored in the memory. When the processor executes the program, it implements the rapid extraction method for the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel as described above.

[0020] The present invention also provides a storage medium storing a program that, when executed, implements the method for rapid extraction of the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel as described above.

[0021] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention first performs noise reduction processing on the three-dimensional laser point cloud of the initial support of the highway tunnel, and then performs noise reduction again by calculating the edge point cloud through two-plane projection and slicing method. This can quickly and accurately remove noise information in the three-dimensional laser point cloud. Finally, by using the coordinates of the central axis point cloud on the two planes and combining the nearest neighbor relationship of the point, the central axis point cloud data of the initial support can be extracted quickly and accurately.

[0022] (2) This invention uses the Algha-Shape algorithm to extract contour points on two planes, through an adjustable parameter. α By controlling the tightness of the contours, the Algha-Shape algorithm can delineate contours that closely match the real shape within the deep point cloud. Furthermore, the algorithm is well-suited for complex surface reconstruction, providing excellent contours for such surfaces.

[0023] (3) The present invention can reflect the position of the central axis of the point cloud in the three-dimensional point cloud more accurately by projecting the point cloud in two directions, thereby reducing the error; and can reflect the tunnel alignment changes to a certain extent, making the acquisition of the central axis of the tunnel point cloud more accurate.

[0024] (4) The planar projection of the present invention converts the three-dimensional point cloud into two two-dimensional point clouds. The slicing algorithm extracts the central axis point cloud on the two projections, which not only has high stability and strong noise resistance, but also obtains a smooth spatial curve as the central axis by calculating the coordinates of the central axis of all slices in sequence, thus avoiding the jagged or discontinuous broken lines generated by the three-dimensional algorithm. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is the preliminary denoising result of the present invention; Figure 3 This is the tunnel initial support point cloud block extraction process of the present invention; Figure 4 This is the result of extracting cloud blocks at the tunnel initial support point according to the present invention; Figure 5 The point cloud projection results of the present invention are shown in Figure (1), which is the point cloud projection result on the XOY plane, and Figure (2) is the point cloud projection result on the XOZ plane. Figure 6The Algha-Shape algorithm of this invention is based on the principle of contouring, wherein (1) Figure shows the principle of forming point set Q, and (2) Figure shows the principle of calculating the coordinates of two circle centers; Figure 7 The results of edge point extraction in this invention are shown in Figure (1), which shows the edge point extraction results on the XOY plane, and Figure (2) shows the edge point extraction results on the XOZ plane. Figure 8 The tunnel centerline extraction results of the present invention are shown in Figure (1) as the centerline on the XOY plane and Figure (2) as the centerline on the XOZ plane. Detailed Implementation

[0026] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0027] Example 1 This embodiment provides a rapid method for extracting the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel, such as... Figure 1 As shown, the method includes the following steps: S1: Obtain point cloud data and perform point cloud denoising: This step, based on on-site measurement benchmarks of the highway tunnel, uses a 3D laser scanner to acquire the 3D laser point cloud of the current tunnel's initial support. The scanned point cloud is then preprocessed using distance filtering and Euclidean clustering to remove noise. Specifically, the steps include: 1. Obtain point cloud and perform preliminary point cloud denoising. (1) Point cloud acquisition By utilizing the layout control points inside the tunnel, a 3D laser scanner station was established, and combined with the resection method, rapid scanning of the initial support point cloud inside the tunnel was achieved.

[0028] (2) Point cloud noise reduction First, preliminary denoising is performed. The point cloud is clustered using the Euclidean distance growing algorithm. The cluster with the largest number of points is considered the initial main body of the point cloud, and other point cloud clusters are removed. The main process is as follows: (1) Mark all points as unclassified, denoted as point set α; (2) Select any 1 point from the point set α Place them into the empty point set Ω and mark them as classified. (3) Calculation points If there are neighboring points within the growth distance r, then add the point to be classified into the set γ and proceed to step four; otherwise, proceed to step five. (4) Take any point from the point set γ Place it into the point set Ω, mark it as classified, and repeat step 3; (5) Take the points in the point set Ω as a cluster, and determine whether the number of points in the cluster is greater than the minimum size S of the cluster. If it is less than S, discard the cluster; if it is greater than S, keep the cluster. (6) Return to step (3) until there are no unclassified points in the point cloud. Output the point with the largest number of points in the final point cloud as the result point cloud to obtain the three-dimensional laser point cloud after preliminary denoising.

[0029] Among these, the most important parameters are the minimum cluster size S and the growth distance r. Preferably, the minimum cluster size S is 1 / k of the total number of point clouds n; generally, k can be 10. The growth distance r is 0.1. The final result is as follows: Figure 2 The preliminary denoising results are shown.

[0030] Although the discrete and isolated points in the initially denoised 3D laser point cloud have been removed, a large number of large-scale noise points still exist due to the complex facilities inside the tunnel, and these points intersect with the initial support. By combining the point cloud normal vector features, the Euclidean clustering algorithm was improved, enabling rapid extraction of the tunnel's initial support.

[0031] 2. Initial branch extraction from point cloud like Figure 3 As shown, the extraction process includes the following: (1) For the three-dimensional laser point cloud after preliminary denoising, the normal vector of each point cloud is calculated based on the PCA (principal component analysis) algorithm; (2) Since the normal vectors of point cloud data are ambiguous, the sign of the dot product is used to determine the ambiguity of the normal vectors and to unify the direction of the normal vectors. The formula is as follows: , In the formula, Let be the normal vector of that point. This is the reference direction vector; the reference direction vector is generally taken as the direction vector between this point and the position of the laser scanning device.

[0032] (3) Use any point as the query point Calculate the query point p i With each point in the neighborhood set The included angle And sum up all the included angles to get Calculate the average included angle. And set the included angle threshold. If the included angle If the value is positive, then the point is considered noise and removed; otherwise, it is not.

[0033] (4) Based on the Euclidean clustering method, the point cloud is divided into blocks and clustered. The point cloud block with the most point cloud is taken as the denoised tunnel initial support point cloud block.

[0034] S2: Project onto the planes in two directions and obtain the point cloud of the plane boundary: This step projects the processed tunnel initial support point cloud block into two planar point clouds along the XOY and XOZ directions, and then uses the Algha-Shape algorithm to obtain the edge point clouds (contours) of the two planar point clouds. Specifically, it includes the following steps: (1) Point cloud planar projection Based on the tunnel initial support point cloud block obtained after point cloud denoising, the point cloud coordinates are transformed from XYZ coordinates to XY and YZ plane coordinates, that is, projected onto the XOY and XOZ directions to obtain the projection results of the point cloud on the XOY and XOZ planes, forming the XOY plane point cloud. Figure 5 (1) Figure and XOZ plane point cloud ( Figure 5 (Figure 2) in the middle.

[0035] (2) Acquisition of planar boundary point cloud To better obtain the outer contour (boundary) of the point cloud, the Algha-Shape algorithm is used to perform edge detection and extraction on the XOY and XOZ plane point clouds. The specific process is as follows: ① For any point The rolling radius is α The circle is searched for within the point cloud to determine the distance. p All points within a fixed threshold, where the threshold value is 2. α Let Q be the point set, such as Figure 6 As shown in Figure (1).

[0036] ② Take any point in the point set Q According to the point and points coordinates and rolling radius Calculate For strings, Coordinates of the centers of two circles with radius and ,like Figure 6 As shown in Figure (2), its coordinates are calculated using the following formula: , , , , In the formula, The radius of the rolling circle, d for .

[0037] (3) In the Q-point set, calculate the division of... p 1 Outside of point o 1 distance L 1 and to o 2 distance L 2 When all points L 1 and L 2 When all values ​​are greater than α, p The point is the boundary point of the point cloud.

[0038] (4) If L 1 and L 2 If not all points are greater than α, then choose another point as... p 1 Point. After calculating the new center coordinates, repeat the calculation process of the previous step and make a judgment. If after traversing all points in the Q neighborhood, no point satisfies the condition of step (3), then p A point is not a boundary point of a point cloud.

[0039] The calculated boundary points of the point cloud are as follows Figure 7 As shown, where Figure 7 Figure (1) shows the projected contour points in the XOY direction, i.e., the boundary point cloud of the XOY plane. Figure 7 Figure (2) shows the projected contour points in the XOZ direction, i.e., the boundary point cloud of the XOZ plane.

[0040] S3: Obtain the point cloud coordinates of the centerlines of the two-directional planes: This step, based on the tunnel projection contour points obtained in step S2, uses the slicing method to calculate and generate the central axis of the tunnel point cloud on two planes. Specifically, it includes the following steps: For the boundary point clouds of the XOY plane and the XOZ plane, given a two-dimensional contour point cloud The method for extracting the tunnel plane centerline based on the slicing method is as follows: (1) Based on the PCA algorithm, calculate the principal direction of point cloud P and the center point C of point cloud to obtain the principal direction vector. and slice direction vector ; (2) Place each point on the outline p i Towards the main direction Projecting upwards, we obtain a scalar. s iThis value represents the position of the point on the principal axis, as shown in the formula below: , After traversing all points on the contour, the distribution range of the point cloud in the main direction can be obtained.

[0041] (3) Set the slice thickness d and slice step size .from s i The search for the slice center begins at the minimum position. For a given slice center position... k Collect all points that meet the following conditions p i The point cloud set that constitutes this tunnel section, among which p i The value requirements are as follows: , Filtered p i Point projection to In the direction, a set of scalars is obtained. t i Sequence, and take the said t i scalars of sites in the sequence t i ’ As the centerline point corresponding to the slice.

[0042] (4) Delete the processed point cloud and continue processing the remaining point cloud according to the slicing method until the point cloud processing is complete. Take the coordinates of all the centerline points processed by the slicing method in the XY coordinate system as the centerline point cloud coordinates on the XOY plane; Based on the above algorithm, the central axis can be calculated as follows: Figure 8 As shown, where Figure 8 Figure (1) shows the central axis obtained on the XOY plane. Figure 8 Figure (2) shows the central axis obtained on the XOZ plane.

[0043] S4: Determine the coordinates of the 3D centerline point cloud based on nearest neighbor relationships: This step combines the point cloud coordinates of the XOY and XOZ axes, uses a two-dimensional axis coordinate as a reference, and calculates the third coordinate based on the nearest neighbor relationship to complete the calculation of the three-dimensional coordinates.

[0044] Since the coordinates of the centerline of a 3D point cloud are generally ( x,y,z The coordinates of the L1 set of points along the XOY plane are in the form of ( ). x,y The coordinates of the L2 set of points along the XOZ plane centerline are ( y,zTherefore, to obtain the 3D point cloud coordinates based on the two planar point cloud coordinates, the following approach should be followed: ① Set the coordinates of the centerline points in the XOY plane to L1 ( x ci ,y ci ) is converted into the coordinates of the central axis L3 of the 3D point cloud, that is ( x ci ,y ci , 0); ②For any L1 y ci Searching for L2 and y ci Given n adjacent points, obtain the corresponding points in L2. z ci Find the value and obtain n values. z ci The average value is used as the z value of the corresponding L3 point cloud to confirm the coordinates of a three-dimensional centerline point cloud. ③ Continue in this manner to confirm the L3 coordinates of all three-dimensional centerline point clouds.

[0045] The above method is based on the C++ platform and PCL point cloud library, and adopts fully automated processing, which is highly efficient and easy to operate.

[0047] Example 2 This embodiment provides a rapid extraction system for the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel, including a memory, a processor, and a program stored in the memory. When the processor executes the program, it implements the rapid extraction method for the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel as described in Embodiment 1 above.

[0048] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0049] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0050] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0051] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0052] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0053] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0054] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for rapid extraction of the centerline of a three-dimensional laser point cloud of the initial support of a highway tunnel, characterized in that, Includes the following steps: The three-dimensional laser point cloud of the initial support of the highway tunnel is obtained, the point cloud is denoised, and the denoised tunnel initial support point cloud block is extracted. The denoised tunnel initial support point cloud block is projected onto the XOY and XOZ planes to form XOY plane point cloud and XOZ plane point cloud, and edge detection is performed to extract XOY plane boundary point cloud and XOZ plane boundary point cloud. Based on the boundary point clouds of the XOY plane and the XOZ plane, the coordinates of the central axis point cloud on the XOY plane and the XOZ plane are calculated using the slicing method. Based on the point cloud coordinates of the central axis on the XOY and XOZ planes, the final three-dimensional point cloud coordinates of the central axis are determined based on the nearest neighbor relationship.

2. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 1, characterized in that, The steps for extracting the denoised tunnel initial support point cloud blocks include: Based on the three-dimensional laser point cloud of the initial support of the highway tunnel, the Euclidean distance growing algorithm is used for preliminary denoising to obtain the three-dimensional laser point cloud after preliminary denoising. Based on the 3D laser point cloud after initial denoising, the PCA algorithm is used for further denoising, and the Euclidean clustering algorithm is combined to extract the denoised tunnel initial support point cloud blocks.

3. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 2, characterized in that, The steps for obtaining the preliminarily denoised 3D laser point cloud include: 1) Mark all points in the three-dimensional laser point cloud as unclassified, denoted as point set α; 2) Take any point from the point set α. Place them into the point set Ω and mark them as classified; 3) Calculation points Are there any neighboring points within the growth distance r? If so, then move the point... If the points are placed into the set γ to be classified, proceed to step 4; otherwise, proceed to step 5. 4) Randomly select a point from the set of points to be classified, γ. Place it into the point set Ω, mark it as classified, and repeat step 3); 5) Take the points in the point set Ω as a cluster, and determine whether the number of points in the cluster is greater than the minimum size S of the cluster. If yes, keep the cluster; otherwise, discard the cluster. The minimum size S of the cluster is taken as 1 / k of the total number of point clouds n, where k is a set constant. 6) Return to step 3) until there are no more unclassified points, and output the cluster with the most points as the 3D laser point cloud after preliminary denoising.

4. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 2, characterized in that, The steps for extracting the denoised tunnel initial support point cloud blocks include: For the preliminarily denoised 3D laser point cloud, the PCA algorithm is used to calculate the normal vector of each point cloud. ; Based on the normal vectors of each point cloud, the sign of the dot product is used to determine the ambiguity of the normal vectors, so as to unify the direction of all normal vectors and obtain point cloud normal vectors with unified direction. The expression for unifying the direction is as follows: ， In the formula, For the first i The normal vector of each point The reference direction vector; Based on the point cloud normal vectors with unified orientation, any point is used as the query point. Calculate the query point With each point in the neighborhood set The included angle And sum up all the included angles to get the total included angle. ; Based on the sum of the included angles Calculate the average included angle And determine whether it satisfies If so, the query point If a point is considered noise, it is removed; otherwise, it is not considered noise and is not removed. This completes the denoising process, resulting in a denoised 3D laser point cloud. The set included angle threshold; For the denoised 3D laser point cloud, Euclidean clustering is used for block clustering, and the point cloud block with the largest number of points is taken as the denoised tunnel initial support point cloud block.

5. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 1, characterized in that, The edge detection was performed using the Algha-Shape algorithm.

6. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 5, characterized in that, The steps for obtaining the XOY plane boundary point cloud and the XOZ plane boundary point cloud include: (1) Take any point in the XOY plane point cloud Centered on, based on the rolling radius α The circle defined by 2 α Distance threshold, search all [distance thresholds] p The distance between the points does not exceed 2 α Let the points be denoted as point set Q; (2) Take any point in the point set Q According to the point and points coordinates and rolling radius Calculate For strings, Coordinates of the centers of two circles with radius and The expression for calculating the center coordinates is: ， ， ， ， In the formula, Where is the rolling radius, ; (3) In the point set Q, calculate except p 1 Coordinates of all points other than the center point to the center of the circle o 1 distance L 1 Sum to the center coordinates o 2 distance L 2 ; (4) Determine when excluding p 1 Distance to all points other than point 1 L 1 and distance L 2 Are they all greater than α ,if, p If a point is selected as the boundary point, then another point is selected from the point set Q. p 1 Return to step (2) to calculate the new distance. L 1 and distance L 2 And re-evaluate; after traversing all points in the point set Q, none of them satisfy the condition greater than 1. α The conditions for judgment p The point is not a boundary point; (5) Denote all boundary points as the boundary point cloud of the XOY plane; For the XOZ plane point cloud, repeat all the above steps to obtain all edge points on the XOZ plane, denoted as the XOZ plane edge point cloud.

7. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 1, characterized in that, The steps for calculating the coordinates of the central axis point cloud on the XOY and XOZ planes include: For the XOY plane boundary point cloud P, calculate its principal direction and center point C to obtain the principal direction vector. and slice direction vector ; Points on the XOY plane boundary point cloud p i To the main direction vector Projecting upwards, we obtain the representation points. p i Scalar of position on the principal axis s i The scalar s i The calculation expression is: ， Set slice thickness d and slice step size ; scalar of minimum value s i As the starting position, at the slice step size Find the set slice center position k To obtain the points that satisfy the constraints. p i The set of point clouds representing tunnel sections, wherein the constraints are: ， The points in the tunnel section point cloud set p i Projected onto the principal direction vector Above, we obtain a set of scalars. t i Sequence, and take the said t i scalars of sites in the sequence t i ’ As the centerline point corresponding to the slice; Delete the processed point cloud and continue to process the remaining point cloud according to the slicing method until the point cloud processing is completed. Finally, take the coordinates of all the centerline points in the XY coordinate system processed by the slicing method as the centerline point cloud coordinates on the XOY plane. For the boundary point cloud of the XOZ plane, repeat all the above steps to obtain the coordinates of the central axis point cloud on the XOZ plane.

8. The method for rapid extraction of the centerline of the three-dimensional laser point cloud of the initial support of a highway tunnel according to claim 1, characterized in that, The steps for determining the final three-dimensional centerline point cloud coordinates include: Define the coordinates of the centerline point cloud on the XOY plane as L1 and the coordinates of the centerline point cloud on the XOZ plane as L2; Find the coordinates of all points in L1. x ci ,y ci Converting ) to the centerline coordinates L3 of a 3D point cloud, denoted as ( x ci ,y ci , 0); Find any point in L2 that is in L1 y ci By taking the values ​​of n adjacent points, obtain the corresponding points in the point cloud coordinates of the central axis on the XOZ plane. z ci Find the value and find the nth one. z ci The average value is used as the coordinate L3 of the central axis of the three-dimensional point cloud. z The value confirms the coordinates of the central axis of a 3D point cloud. For the remaining points in L1, repeat all the above steps to confirm the coordinates of the central axis of all three-dimensional point clouds in L3, obtain the final coordinates of the three-dimensional central axis point cloud, and fit a straight line to obtain the central axis of the tunnel's initial support.

9. A rapid extraction system for the centerline of a three-dimensional laser point cloud of the initial support of a highway tunnel, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method for rapid extraction of the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel as described in any one of claims 1-8.

10. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the method for rapid extraction of the central axis of the three-dimensional laser point cloud of the initial support of a highway tunnel as described in any one of claims 1-8.

Images