Tunnel section fitting method and device based on laser radar point cloud and medium

Abstract

The invention discloses a tunnel section fitting method and device based on a laser radar point cloud, and a medium, and the method comprises the steps: scanning a subway tunnel through employing a laser radar, and obtaining the point cloud data of the subway tunnel; obtaining a tunnel trend by using a principal component analysis method, and performing coordinate transformation on the point cloud; obtaining point cloud slices along the tunnel trend; projecting the slice point cloud to a YOZ plane, constructing a polar coordinate system by taking the mass center of the slice point cloud as a circle center, and converting slice points to polar coordinates; uniformly dividing the section point cloud according to polar coordinate angles; and improving a random sampling consistency algorithm by using angle distribution, and realizing elliptic model fitting of the section point cloud. According to the method, the improved random sampling consistency algorithm based on polar coordinates is introduced, the problems that the section ellipse model is low in fitting efficiency, prone to falling into local optimum and the like are solved, the precision, reliability and efficiency of tunnel section fitting are improved, and optimal fitting of the tunnel section ellipse model is achieved.

Description

Technical Field

[0001] The present invention belongs to a processing technology based on laser radar point cloud data, and specifically relates to a tunnel section fitting method, device and medium based on laser radar point cloud, and is particularly applied to subway tunnel detection. Background Art

[0002] Tunnels are an important part of the transportation network, and their structural safety is crucial to ensuring smooth traffic and passenger safety. With the development of technology, lidar has been widely used in tunnel detection, providing high-precision three-dimensional point cloud data. However, traditional point cloud processing methods usually treat tunnel cross-sections as circles. However, due to the influence of construction, deformation, etc., the tunnel cross-section in operation is not a circle, but closer to an ellipse. If it is treated as a circle, it will directly lead to the convergence, displacement and other information obtained through the cross-section being inconsistent with the actual situation, affecting the authenticity of the detection. The present invention optimizes the random sampling consistency algorithm by constructing a polar coordinate optimization sample point selection process, thereby efficiently performing high-precision fitting of the tunnel cross-section ellipse model, providing reliable technical support for subway operation and maintenance, and will bring more efficient and reliable solutions to subway operation management. Summary of the invention

[0003] Purpose of the invention: The present invention aims to provide a tunnel section fitting method, device and medium based on lidar point cloud, and adopts an improved RANSAC algorithm to realize the fitting of the tunnel section ellipse model, so that the detection result is more realistic.

[0004] Technical solution: The tunnel section fitting method based on laser radar point cloud described in the present invention comprises the following steps:

[0005] (1) Use laser radar to scan subway tunnels and obtain original point cloud data of subway tunnels. The observation values ​​are the three-dimensional coordinates of the surface points inside the tunnel;

[0006] (2) Use the principal component analysis method to obtain the tunnel direction and perform coordinate transformation on the point cloud;

[0007] (3) Obtain point cloud slices along the tunnel direction;

[0008] (4) Project the slice point cloud onto the YOZ plane, construct a polar coordinate system with the centroid of the slice point cloud as the center of the circle, and convert the slice points into polar coordinates;

[0009] (5) Divide the cross-section point cloud evenly according to the polar coordinate angle;

[0010] (6) The random sampling consistency algorithm is improved by using the angle distribution to realize the elliptical model fitting of the cross-section point cloud.

[0011] Furthermore, the implementation process of step (1) is as follows:

[0012] The original point cloud set is P(x i ,y i ,z i ) i＝1,2,…,n , construct the covariance matrix for its XY coordinates:

[0013]

[0014] Where n is the number of points in the collected point cloud, are the coordinates of the center of gravity of the point set plane.

[0015] Furthermore, the implementation process of step (2) is as follows:

[0016] Performing principal component analysis on the matrix C, we get two eigenvalues ​​λ1 and λ2, λ1≥λ2>0, and the corresponding two eigenvectors are v1 and v2, where v1 represents the direction of the tunnel. Then, the coordinate transformation process is as follows:

[0017] Q=P[v1,v2,(0,0,1) T ]

[0018] After coordinate transformation, the tunnel point cloud coordinate Q is obtained, where the X-axis is parallel to the tunnel direction, the Y-axis is perpendicular to the tunnel direction, and the Z-axis is consistent with the Z-axis of the original point cloud coordinate.

[0019] Furthermore, the implementation process of step (3) is as follows:

[0020] The thickness of the point cloud slice is w, and the maximum and minimum values ​​of the x coordinate in the coordinate Q are calculated, which are x max and x min , then the tunnel point cloud is divided into nslice slices, and nslice is calculated by the following formula:

[0021] nslice=floor((x max -x min ) / w)

[0022] Among them, floor(·) means rounding up, and the coordinates of the point cloud in each slice are

[0023] Furthermore, the implementation process of step (4) is as follows:

[0024] The centroid of the sliced ​​point cloud is avg_p:

[0025]

[0026] Among them, m k is the number of points in the kth slice;

[0027] With the centroid as the center of the circle, the points inside the slice are transformed into the polar coordinate system, specifically:

[0028]

[0029]

[0030] Then the polar coordinates of the slice point cloud are

[0031] Furthermore, the implementation process of step (5) is as follows:

[0032] The angle distribution range of the cross-section point cloud in the polar coordinate system is 0° to 360°. The cross-section points are divided into floor(360 / Δθ) areas according to the angle interval of Δθ.

[0033] Furthermore, the implementation process of step (6) is as follows:

[0034] (61) Input the 2D coordinates of the cross section Data, the general equation of an ellipse is:

[0035] f(y,z)=Ay 2 +Byz+Cz 2 +Dy+Ez+F=0

[0036] In the formula, A, B, C, D, E, and F are unknown coefficients, A and C are not zero, and B 2 -4AC>0;

[0037] (62) A certain number of points are selected with equal probability in each region for ellipse fitting;

[0038] (63) Calculate the error between the model and each point in the slice data, set the threshold to s, and mark the points within the allowable error threshold as “inliers”;

[0039] (64) Compare the ellipse model with the model saved in the previous iteration, and save the “inside point” data with higher accuracy;

[0040] (65) Repeat steps (62)-(64) until the optimal ellipse model is obtained and the iteration ends;

[0041] (66) The optimal “inside point” data were processed using the least squares ellipse fitting method to obtain the best model.

[0042] Furthermore, according to the tunnel section fitting method based on laser radar point cloud according to claim 6, it is characterized in that the Δθ is 60°.

[0043] A device according to the present invention includes a memory and a processor, wherein:

[0044] A memory for storing computer programs that can be run on the processor;

[0045] The processor is used to execute the steps of the above-mentioned tunnel section fitting method based on laser radar point cloud when running the computer program.

[0046] A storage medium described in the present invention stores a computer program on the storage medium, and when the computer program is executed by at least one processor, the steps of the tunnel section fitting method based on laser radar point cloud as described above are implemented.

[0047] Beneficial effects: Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention breaks through the existing tunnel section circle fitting method, and solves the impact of tunnel deformation during operation on data processing by adopting an ellipse fitting method; 2. The present invention has high accuracy and efficiency, and fully utilizes the advantages of dense point clouds of lidar technology. Under the premise of ensuring accuracy, it ensures the efficient fitting of the ellipse model; 3. The method proposed by the present invention improves the basic data accuracy for the later monitoring of subway tunnel deformation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of the tunnel section fitting method based on laser radar point cloud of the present invention;

[0049] Figure 2 A subway tunnel point cloud scatter plot obtained for the implementation of the present invention;

[0050] Figure 3 The tunnel point cloud scatter diagram after the coordinate conversion is implemented in the present invention;

[0051] Figure 4 The tunnel point cloud scatter diagram after slicing is implemented in the present invention;

[0052] Figure 5 A schematic diagram of converting slice points to polar coordinates according to the present invention;

[0053] Figure 6 This is a schematic diagram of evenly dividing the cross-section point cloud according to polar coordinate angles in the present invention;

[0054] Figure 7 The schematic diagram of the improved RANSAC algorithm for obtaining the optimal ellipse fitting model of the cross section is implemented in the present invention. DETAILED DESCRIPTION

[0055] The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0056] like Figure 1 As shown, the present invention provides a tunnel section fitting method based on laser radar point cloud, comprising the following steps:

[0057] Step 1: If Figure 2 As shown in the figure, the laser radar scans the subway tunnel and obtains the subway tunnel point cloud data P(x i ,y i ,z i ) i＝1,2,…,n , where n is the number of points in the collected point cloud; the observation value is the three-dimensional coordinate of the surface point inside the tunnel.

[0058] The original point cloud set is P(x i ,y i ,z i ) i＝1,2,…,n , construct the covariance matrix for its XY coordinates:

[0059]

[0060] in, are the coordinates of the center of gravity of the point set plane.

[0061] Step 2: Use the principal component analysis method to obtain the tunnel direction and perform coordinate transformation on the point cloud.

[0062] By performing principal component analysis on the matrix C, we can obtain two eigenvalues ​​λ1 and λ2 (λ1≥λ2>0), and the corresponding two eigenvectors are v1 and v2, where v1 represents the direction of the tunnel. Then, the coordinate transformation process is as follows:

[0063] Q=P[v1,v2,(0,0,1) T ]

[0064] After coordinate transformation, the tunnel point cloud coordinates Q are obtained. The tunnel point cloud after coordinate transformation is as follows: Figure 3 As shown, the X-axis is parallel to the tunnel direction, the Y-axis is perpendicular to the tunnel direction, and the Z-axis is consistent with the Z-axis of the original point cloud coordinates.

[0065] Step 3: Obtain point cloud slices along the tunnel direction.

[0066] The point cloud slice thickness w is set to 1.2m, and the maximum and minimum values ​​of the x coordinate in the coordinate Q are calculated, which are x max and x min , then the tunnel point cloud can be divided into 21 slices, and the number of slices nslice can be calculated by the following formula:

[0067] nslice=floor((x max -x min ) / w)

[0068] Among them, floor(·) means rounding up, and the coordinates of the point cloud in each slice are The sliced ​​tunnel point cloud is as follows Figure 4 shown.

[0069] Step 4: Figure 5 As shown, the slice point cloud is projected onto the YOZ plane, a polar coordinate system is constructed with the centroid of the slice point cloud as the center of the circle, and the slice points are converted to polar coordinates. The details are as follows:

[0070] Calculate the centroid of the slice point cloud as avg_p:

[0071]

[0072] Among them, m k is the number of points in the kth slice.

[0073] With the centroid as the center of the circle, the points inside the slice are transformed into the polar coordinate system, specifically:

[0074]

[0075] Then the polar coordinates of the slice point cloud are

[0076] Step 5: Divide the cross-section point cloud evenly according to the polar coordinate angles.

[0077] The angle distribution range of the cross-section point cloud in the polar coordinate system is 0° to 360°. The cross-section points are divided into floor(360 / Δθ) areas according to the angle interval of Δθ. Here, Δθ is set to 60°, that is, the cross-section points are divided into 6 parts according to the angle interval, such as Figure 6 shown.

[0078] Step 6: Figure 7 As shown in the figure, the RANdom SAmpleConsensus (RANSAC) algorithm is improved by using the angle distribution to realize the ellipse model fitting of the cross-section point cloud. The details are as follows:

[0079] 1) Input the 2D coordinates of the cross section Data, the general equation of an ellipse is:

[0080] f(y,z)=Ay 2 +Byz+Cz 2 +Dy+Ez+F=0

[0081] In the formula, A, B, C, D, E, and F are unknown coefficients, A and C are not zero, and B 2 -4AC>0. For the equation of an ellipse, at least five points are required to determine the curve.

[0082] 2) In the area divided in step 5, a certain number of points are equally likely selected from the five areas IV for ellipse fitting. Considering that the area directly below the cross section is the track and road surface points, these points deviate from the ellipse model. Therefore, these points are not randomly sampled in the model optimization, which can improve the accuracy and efficiency of the model.

[0083] 3) Calculate the error between the model and each point in the slice data, set the threshold to s, and mark the points within the allowable error threshold as "inliers".

[0084] 4) Compare the model with the model saved in the previous iteration and save the "inlier" data with higher accuracy.

[0085] 5) Repeat steps 2)-4) until the optimal ellipse model is obtained and the iteration ends.

[0086] 6) The optimal "inside point" data is processed using the least squares ellipse fitting method to obtain the best model.

[0087] The present invention also provides a device, including a memory and a processor, wherein: the memory is used to store a computer program that can be run on the processor; the processor is used to execute the steps of the above-mentioned tunnel section fitting method based on laser radar point cloud when running the computer program.

[0088] The present invention also provides a storage medium having a computer program stored thereon. When the computer program is executed by at least one processor, the steps of the tunnel section fitting method based on laser radar point cloud as described above are implemented.

[0089] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A tunnel section fitting method based on laser radar point cloud, characterized in that: The following steps are involved: (1) Use laser radar to scan subway tunnels and obtain original point cloud data of subway tunnels. The observation values ​​are the three-dimensional coordinates of the surface points inside the tunnel; (2) Use the principal component analysis method to obtain the tunnel direction and perform coordinate transformation on the point cloud; (3) Obtain point cloud slices along the tunnel direction; (4) Project the slice point cloud onto the YOZ plane, construct a polar coordinate system with the centroid of the slice point cloud as the center of the circle, and convert the slice points into polar coordinates; (5) Divide the cross-section point cloud evenly according to the polar coordinate angle; (6) The random sampling consistency algorithm is improved by using the angle distribution to realize the elliptical model fitting of the cross-section point cloud.

2. The tunnel section fitting method based on laser radar point cloud according to claim 1 is characterized in that: The implementation process of step (1) is as follows: The original point cloud set is P(x i ,y i ,z i ) i＝1,2,…,n , construct the covariance matrix for its XY coordinates: Where n is the number of points in the collected point cloud, are the coordinates of the center of gravity of the point set plane.

3. The tunnel section fitting method based on laser radar point cloud according to claim 1 is characterized in that: The implementation process of step (2) is as follows: Performing principal component analysis on the matrix C, we get two eigenvalues ​​λ1 and λ2, λ1≥λ2>0, and the corresponding two eigenvectors are v1 and v2, where v1 represents the direction of the tunnel. Then, the coordinate transformation process is as follows: Q=P[v1,v2,(0,0,1) T ] After coordinate transformation, the tunnel point cloud coordinate Q is obtained, where the X-axis is parallel to the tunnel direction, the Y-axis is perpendicular to the tunnel direction, and the Z-axis is consistent with the Z-axis of the original point cloud coordinate.

4. The tunnel section fitting method based on laser radar point cloud according to claim 1, characterized in that: The implementation process of step (3) is as follows: The thickness of the point cloud slice is w, and the maximum and minimum values ​​of the x coordinate in the coordinate Q are calculated, which are x max and x min , then the tunnel point cloud is divided into nslice slices, and nslice is calculated by the following formula: nslice=floor((x max -x min ) / w) Among them, floor(·) means rounding up, and the coordinates of the point cloud in each slice are k=1,2,…,nslice.

5. The tunnel section fitting method based on laser radar point cloud according to claim 1, characterized in that: The implementation process of step (4) is as follows: The centroid of the sliced ​​point cloud is avg_p: Among them, m k is the number of points in the kth slice; With the centroid as the center of the circle, the points inside the slice are transformed into the polar coordinate system, specifically: Then the polar coordinates of the slice point cloud are 6. The tunnel section fitting method based on laser radar point cloud according to claim 1, characterized in that: The implementation process of step (5) is as follows: The angle distribution range of the cross-section point cloud in the polar coordinate system is 0° to 360°. The cross-section points are divided into floor(360 / Δθ) areas according to the angle interval of Δθ.

7. The tunnel section fitting method based on laser radar point cloud according to claim 1, characterized in that: The implementation process of step (6) is as follows: (61) Input the 2D coordinates of the cross section Data, the general equation of an ellipse is: f(y,z)=Month 2 +Byz+Cz 2 +Dy+Ez+F=0 In the formula, A, B, C, D, E, and F are unknown coefficients, A and C are not zero, and B 2 -4AC>0; (62) A certain number of points are selected with equal probability in each region for ellipse fitting; (63) Calculate the error between the model and each point in the slice data, set the threshold to s, and mark the points within the allowable error threshold as "inliers"; (64) Compare the ellipse model with the model saved in the previous iteration, and save the "inside point" data with higher accuracy; (65) Repeat steps (62)-(64) until the optimal ellipse model is obtained and the iteration ends; (66) The optimal “inside point” data were processed using the least squares ellipse fitting method to obtain the best model.

8. The tunnel section fitting method based on laser radar point cloud according to claim 6, characterized in that: The Δθ is 60°.

9. A device, characterized in that: comprising a memory and a processor, wherein: A memory for storing computer programs that can be run on the processor; A processor is used to execute the steps of a tunnel section fitting method based on laser radar point cloud as described in any one of claims 1 to 8 when running the computer program.

10. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by at least one processor, implements the steps of a tunnel section fitting method based on a laser radar point cloud as described in any one of claims 1 to 8.

Images