High-precision tunnel three-dimensional reconstruction method and reconstruction device based on CPⅢ network

Abstract

The application discloses a high-precision tunnel three-dimensional reconstruction method and device based on a CP III network, and the reconstruction method comprises the following steps: S1, establishing a CP III network, and making the coordinates of each base mark known under the CP III network, and numbering each base mark; S2, providing a holder trolley, and making the holder trolley drive along a tunnel to acquire point cloud data of the tunnel, detection coordinates of each base mark and a detection running track of the holder trolley; S3, sticking the acquired detection coordinates of the base mark into a coordinate system of the CP III network, and comparing the detection coordinates of the base mark with the coordinates of the control points, correcting the detection running track section at the base mark where an error is found, and correcting corresponding point cloud data; and S4, generating a three-dimensional model of the tunnel according to the point cloud data. The control points of the CP III network with known coordinates in the tunnel are taken as reference points, the detected tunnel holder data is corrected, more accurate tunnel holder data is obtained, and the modeling precision of the three-dimensional reconstruction of the tunnel is improved.

Description

Technical Field

[0001] This invention belongs to the field of building construction, specifically relating to a high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks. Background Technology

[0003] In current tunnel 3D reconstruction work, data acquisition is typically carried out using three methods: traditional geometric and visual methods, deep learning methods, and multi-sensor fusion and positioning optimization.

[0004] Existing technologies employ SLAM (Simultaneous Localization and Mapping) LiDAR (Light Detection and Ranging) for tunnel measurement. This requires measuring the LiDAR's trajectory during the scanning process using an IMU (Installation Unit). However, the gyroscope drift of the IMU is approximately 0.01 gyroscope drift, which can lead to cumulative drift in long tunnels. This results in a significant error between the LiDAR's measured path and its actual path, leading to errors in the point cloud data. Consequently, the accuracy of the tunnel's 3D modeling is insufficient and cannot meet the requirements for subway track laying. Summary of the Invention

[0006] This invention discloses a high-precision 3D reconstruction method and reconstruction equipment for tunnels based on CPⅢ networks. It uses control points of the CPⅢ network with known coordinates within the tunnel as reference points to correct the tunnel PTZ data, thereby obtaining more accurate tunnel PTZ data and improving the modeling accuracy of tunnel 3D reconstruction.

[0007] This invention discloses a high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks, comprising the following steps: S1. Set up multiple control points in the tunnel and set up a benchmark at the location of each control point. Establish a CPⅢ network based on the tunnel, and make the coordinates of each benchmark known in the CPⅢ network, and number each benchmark. S2. Provide a gimbal vehicle equipped with a 3D laser scanner, a depth camera and a trajectory detector, so that the gimbal vehicle travels along the tunnel, and during the travel, the 3D laser scanner, the depth camera and the trajectory detector acquire the point cloud data of the tunnel, the detection coordinates of each of the base points and the detection trajectory of the gimbal vehicle. S3. The obtained detection coordinates of the benchmarks are attached to the coordinate system of the CPⅢ network, and the detection coordinates of each benchmark are compared with the corresponding known coordinates one by one. When it is found that the error between the detection coordinates of a benchmark and the corresponding known coordinates exceeds the error threshold, the detection trajectory segment at the benchmark where the error is found is corrected, and the corresponding point cloud data is corrected according to the corrected detection trajectory. S4. Generate a 3D model of the tunnel based on the corrected point cloud data.

[0008] A further improvement of the present invention is that: after performing step S3 and before performing step S4, the corrected point cloud data is preprocessed; when performing step S4, a three-dimensional model of the tunnel is generated based on the preprocessed point cloud data.

[0009] A further improvement of the present invention is that the preprocessing step of the point cloud data includes: The first step is to filter the corrected point cloud data to remove noise points. The second step is to thin out the filtered point cloud data.

[0010] A further improvement of the present invention is that, in step S4, the step of generating a three-dimensional model of the tunnel based on the corrected point cloud data includes: By generating a tunnel triangular mesh model through Poisson reconstruction, discrete point clouds are transformed into continuous curved surfaces.

[0011] This invention discloses a reconstruction device for implementing a high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks, comprising: Multiple base markers are installed on the inner wall of the tunnel and their positions coincide with those of multiple control points. A gimbal trolley can travel along a tunnel. The gimbal trolley is equipped with: a 3D laser scanner for acquiring point cloud data of the tunnel during the travel of the gimbal trolley, a depth camera for acquiring the detection coordinates of each of the base points, and a trajectory detector for acquiring the detection trajectory. The data processing center receives point cloud data, detection coordinates of each benchmark, and detection trajectory acquired by the gimbal vehicle, corrects the point cloud data based on the above data, and generates a three-dimensional model of the tunnel based on the corrected point cloud data.

[0012] A further improvement of the present invention is that a lifting platform is fixed on the gimbal trolley, and the three-dimensional laser scanner is mounted on the lifting platform.

[0013] A further improvement of the present invention is that the gimbal trolley is equipped with an LED fill light module.

[0014] A further improvement of the present invention is that the base is made of a highly reflective material.

[0015] A further improvement of the present invention is that the trajectory detector comprises: An IMU unit used to measure the three-axis angular velocity and three-axis acceleration of the gimbal vehicle in real time in order to calculate its attitude and motion changes; A wheeled odometer for obtaining the relative displacement of the vehicle.

[0016] This invention discloses a high-precision 3D reconstruction method and reconstruction equipment for tunnels based on CPⅢ networks. It uses control points of the CPⅢ network with known coordinates within the tunnel as reference points to correct the detected tunnel PTZ data, thereby obtaining more accurate tunnel PTZ data and improving the modeling accuracy of tunnel 3D reconstruction. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a top view of the gimbal trolley of the present invention in motion; Figure 3 This is a three-dimensional view of the gimbal vehicle of the present invention in motion; Figure 4 This is a schematic diagram of the gimbal trolley structure of the present invention; In the diagram: 1. Tunnel; 2. Base marker; 3. Cruise path; 4. Gimbal trolley; 401. Vehicle body; 402. Lifting platform; 403. Headlights; 404. LED supplementary lighting module; 405. Wheel odometer; 406. Depth camera; 407. 3D laser scanner; 408. IMU unit. Detailed Implementation

[0020] like Figures 1-3 As shown, the present invention provides a high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks, characterized by comprising the following steps: S1. Set up multiple control points in tunnel 1, and set up a benchmark 2 at the location of each control point. Establish a CPⅢ network based on tunnel 1, and make the coordinates of each benchmark 2 known in the CPⅢ network, and number each benchmark 2. S2. Provide a gimbal vehicle 4 equipped with a 3D laser scanner 407, a depth camera 406 and a trajectory detector, so that the gimbal vehicle 4 can travel along the tunnel 1 and obtain the point cloud data of the tunnel 1, the detection coordinates of each benchmark 2 and the detection trajectory of the gimbal vehicle 4 through the 3D laser scanner 407, the depth camera 406 and the trajectory detector during the travel. S3. Fit the obtained detection coordinates of the datum 2 into the coordinate system of the CPⅢ network, and compare the detection coordinates of each datum 2 with the corresponding known coordinates one by one. When the error between the detection coordinates of the datum 2 and the corresponding known coordinates exceeds the error threshold, the detection trajectory segment at the datum where the error was found is corrected, and the corresponding point cloud data is corrected according to the corrected detection trajectory. S4. Generate a 3D model of the tunnel based on the corrected point cloud data.

[0021] Preferably, in this embodiment, the trajectory segment includes a trajectory segment before and after each landmark 2, and ensures that there is a 30% overlap between adjacent trajectory segments.

[0022] like Figure 2 , Figure 3 As shown, in this embodiment, the gimbal vehicle 4 travels along the cruise path 3 shown in the figure. The gimbal vehicle 4 travels along the cruise path 3 on one side of the tunnel 1, and after completing this journey, it travels along the cruise path 3 on the other side of the tunnel 1.

[0023] Preferably, in this embodiment, after the point cloud data is corrected in step S3, the corrected point cloud data is preprocessed before step S4 is executed, and a three-dimensional model of the tunnel is generated based on the preprocessed point cloud data when step S4 is executed.

[0024] Preferably, in this embodiment, the preprocessing of point cloud data specifically includes the following steps: The first step is to filter the corrected point cloud data to remove noise points. The second step is to thin out the filtered point cloud data.

[0025] Preprocessing point cloud data can improve the quality of the subsequently generated 3D tunnel model.

[0026] Preferably, in this embodiment, the step of generating a 3D model of the tunnel based on the corrected point cloud data includes: A triangular mesh model of the tunnel is generated through Poisson reconstruction, transforming the discrete point cloud into a continuous curved surface. Based on the original tunnel design model, the point cloud is fitted onto the parametric model to generate an accurate model. Preferably, in this embodiment, in step S2, point cloud data is obtained by using a laser SLAM algorithm.

[0027] Preferably, in this embodiment, when correcting the point cloud data in step S3, a graph optimization algorithm model is established to transform it into a nonlinear least squares problem for solution. The detection path of the gimbal vehicle 4 consists of pose nodes at multiple moments. The gimbal vehicle 4 treats each pose node as a graph node, and the relative transformation constraints between adjacent pose nodes generated by the IMU unit 408 and the wheeled odometer 405 are used as odometer edge constraints. These constraints have uncertainties (covariance matrix). When the gimbal vehicle 4 passes through the same area again, the SLAM algorithm identifies this as a "closed loop" and generates a strong constraint between pose nodes (closed loop detection edge) to correct drift. Each CPⅢ observation constitutes a CPⅢ observation edge connecting the "vehicle observation pose node" and the "fixed CPⅢ coordinate node". The theoretical value of the edge is the known coordinate of the CPⅢ point, which is the absolute constraint that "pulls" the system back to the real coordinate system. The objective function at this time is to maximize the joint probability of all observations (edges), that is, to minimize the sum of squared residuals (Mahanobis distance) of all constraints: where: It represents all pose nodes that need to be optimized.

[0028] It is the residual of the edge, that is, the difference between the "predicted observation" and the "actual observation".

[0029] For edge CPⅢ, the residual = (coordinates of CPⅢ predicted based on the optimized car pose) - (the known coordinates of CPⅢ).

[0030] Preferably, in this embodiment, as Figure 1 As shown, after establishing the three-dimensional model of the tunnel, the generated three-dimensional model is reviewed, and a cross-sectional deformation analysis report and a foundation construction error report are generated.

[0031] like Figures 2-4 As shown, a reconstruction device for a high-precision 3D reconstruction method in tunnels based on CPⅢ networks includes: Multiple benchmarks are installed on the inner wall of the tunnel, and their positions coincide with those of multiple control points. The gimbal trolley 4 can travel along the tunnel. The gimbal trolley 4 is equipped with a 3D laser scanner 407 for acquiring point cloud data of the tunnel during the travel of the gimbal trolley 4. Figure 1 The LiDAR module in the image, a depth camera 406 for acquiring the detection coordinates of the reference 2, and a trajectory detector for acquiring the detection trajectory; The data processing center receives point cloud data acquired by the gimbal vehicle 4, the detection coordinates of the benchmark 2, and the detection trajectory. Based on the above data, it corrects the point cloud data and generates a three-dimensional model of the tunnel based on the corrected point cloud data.

[0032] like Figure 4 As shown in this embodiment, a lifting platform 402 is fixed on the gimbal trolley 4, and a 3D laser scanner 407 is mounted on the lifting platform 402. By setting the lifting platform 402, the height of the 3D laser scanner 407 can be adjusted according to the actual situation, thereby obtaining more accurate point cloud data.

[0033] like Figure 4 As shown, the gimbal trolley 4 is equipped with an LED supplementary lighting module 404. By setting up the LED supplementary lighting module 404, supplementary lighting can be provided during the movement of the gimbal trolley 4. In tunnel 1 with poor lighting conditions, this can significantly improve the shooting quality of the depth camera 406, thereby obtaining more accurate detection coordinates of the benchmark 2. like Figure 4 As shown, in this embodiment, the gimbal vehicle 4 is equipped with a headlight 403 at the front.

[0034] Preferably, the datum 2 is made of a highly reflective material, making it easier for the depth camera 406 to capture it.

[0035] like Figure 4 As shown, in this embodiment, the trajectory detector includes: IMU unit 408 is used to measure the three-axis angular velocity and three-axis acceleration of the gimbal trolley 44 in real time to calculate its attitude (pitch, roll, yaw) and motion changes. Wheel odometer 405 for obtaining the relative displacement of the trolley.

[0036] The vehicle's pose information is obtained through IMU unit 408, and the vehicle's travel distance is obtained through wheel odometer 405. Combining the two information yields the vehicle's detected travel trajectory.

[0037] This invention can obtain a high-precision 3D tunnel model, but the final accuracy depends on sensor performance, CPⅢ density (±1mm), and the quality of the algorithm. The accuracy of the CPⅢ control network itself is the theoretical upper limit of the overall system accuracy. In subway and high-speed railway tunnels, the relative accuracy of CPⅢ points is usually better than ±1mm (between adjacent points), and the global absolute accuracy is also in the millimeter range. The sensor measurement accuracy is mainly determined by the accuracy of the 3D laser scanner 407 and the IMU unit 408. The ranging accuracy of the 3D laser scanner 407 is usually in the centimeter range (e.g., ±2cm), but its angular resolution is very high, achieving a point accuracy of 3-5mm on the surface of an object at close range (within 10m). The gyroscope drift in the IMU unit 408 is approximately 0.01° / h, which is the main factor causing trajectory drift. Without CPⅢ correction, drift accumulates rapidly.

[0038] The present invention has the following beneficial effects: 1. After rigorous correction using the CPⅢ constraints of this invention, errors caused by the drift of IMU element 408 can be significantly suppressed. Accuracy (positional accuracy of the point cloud in the global coordinate system) can reach ±5mm level due to the constraint of the CPⅢ absolute coordinates. Relative accuracy (accuracy within the point cloud, especially between adjacent points) is even higher, reaching ±2-3mm level. This means that the model can clearly reflect millimeter-level cracks or deformations in the tunnel wall, and has extremely positive significance for track laying construction. 2. This invention eliminates the need for traditional manual station setting and aiming inside the tunnel. Through the automatic movement of the gimbal trolley 408, it achieves automatic detection, accurate extraction and coordinate verification of CPⅢ control points, realizing full automation of point cloud data from acquisition to correction. This completely changes the traditional operation mode that relies on a lot of manual intervention, greatly saves labor costs, and completely avoids the traffic safety risks caused by frequent movement of personnel inside the tunnel. 3. This invention utilizes a high-precision CPⅢ control network as a global absolute constraint for automated correction and evaluation, reducing the data acquisition and processing time that originally took several days to just a few hours. This greatly improves the efficiency of tunnel 3D data acquisition and modeling, ensures the progress requirements of engineering inspection and maintenance, and provides technical support for the rapid digitalization and intelligent management of tunnel structures.

[0039] Finally, it should be noted that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks, characterized in that, Includes the following steps: S1. Set up multiple control points in the tunnel and set up a benchmark at the location of each control point. Establish a CPⅢ network based on the tunnel, and make the coordinates of each benchmark known in the CPⅢ network, and number each benchmark. S2. Provide a gimbal vehicle equipped with a 3D laser scanner, a depth camera and a trajectory detector, so that the gimbal vehicle travels along the tunnel, and during the travel, the 3D laser scanner, the depth camera and the trajectory detector acquire the point cloud data of the tunnel, the detection coordinates of each of the base points and the detection trajectory of the gimbal vehicle. S3. The obtained detection coordinates of the benchmarks are attached to the coordinate system of the CPⅢ network, and the detection coordinates of each benchmark are compared with the corresponding known coordinates one by one. When it is found that the error between the detection coordinates of a benchmark and the corresponding known coordinates exceeds the error threshold, the detection trajectory segment at the benchmark where the error is found is corrected, and the corresponding point cloud data is corrected according to the corrected detection trajectory. S4. Generate a 3D model of the tunnel based on the corrected point cloud data.

2. The high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks as described in claim 1, characterized in that: After step S3 is completed and before step S4 is completed, the corrected point cloud data is preprocessed; when step S4 is completed, a three-dimensional model of the tunnel is generated based on the preprocessed point cloud data.

3. The high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks as described in claim 2, characterized in that, The preprocessing steps for the point cloud data include: The first step is to filter the corrected point cloud data to remove noise points. The second step is to thin out the filtered point cloud data.

4. The high-precision three-dimensional reconstruction method for tunnels based on CPⅢ networks as described in claim 1, characterized in that, In step S4, the step of generating a 3D model of the tunnel based on the corrected point cloud data includes: By generating a tunnel triangular mesh model through Poisson reconstruction, discrete point clouds are transformed into continuous curved surfaces.

5. A reconstruction device for implementing the high-precision tunnel three-dimensional reconstruction method based on CPⅢ network as described in claim 1, characterized in that, include: Multiple base markers are installed on the inner wall of the tunnel and their positions coincide with those of multiple control points. A gimbal trolley can travel along a tunnel. The gimbal trolley is equipped with: a 3D laser scanner for acquiring point cloud data of the tunnel during the travel of the gimbal trolley, a depth camera for acquiring the detection coordinates of each of the base points, and a trajectory detector for acquiring the detection trajectory. The data processing center receives point cloud data, detection coordinates of each benchmark, and detection trajectory acquired by the gimbal vehicle, corrects the point cloud data based on the above data, and generates a three-dimensional model of the tunnel based on the corrected point cloud data.

6. The reconstruction equipment for the high-precision tunnel 3D reconstruction method based on CPⅢ network as described in claim 5, characterized in that, The gimbal trolley is equipped with a lifting platform, and the 3D laser scanner is mounted on the lifting platform.

7. The reconstruction equipment for the high-precision tunnel 3D reconstruction method based on CPⅢ network as described in claim 5, characterized in that, The gimbal trolley is equipped with an LED fill light module.

8. The reconstruction equipment for the high-precision tunnel 3D reconstruction method based on CPⅢ network as described in claim 5, characterized in that, The baseplate is made of a highly reflective material.

9. The reconstruction equipment for the high-precision tunnel 3D reconstruction method based on CPⅢ network as described in claim 5, characterized in that, The trajectory detector includes: An IMU unit used to measure the three-axis angular velocity and three-axis acceleration of the gimbal vehicle in real time in order to calculate its attitude and motion changes; A wheeled odometer for obtaining the relative displacement of the vehicle.

Images