Tunnel deformation automatic detection method and system

Abstract

The invention discloses an automatic tunnel deformation detection method and system, relates to the technical field of tunnel deformation detection, and solves the problem that meaningful feature points and detail deformation are easily ignored in order to simplify the calculation amount in the current tunnel deformation detection, and the method comprises the steps: obtaining tunnel point cloud data of a plurality of intervals, and carrying out the preprocessing; performing registration splicing on the preprocessed tunnel point cloud data of the multiple intervals to obtain complete tunnel point cloud data; slicing the complete tunnel point cloud data to obtain multiple pieces of slice point cloud data; filtering the slice point cloud data based on an improved filtering algorithm; fitting the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; based on the fitting section of the point cloud data of the plurality of slices, analyzing deformation conditions of the whole and details of the tunnel; through filtering processing, meaningful feature points are reserved while the operand is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel deformation detection, and more specifically, to a tunnel deformation automatic detection method and system. Background Art

[0002] Tunnel deformation detection is crucial to ensure the safety, stability and continuity of urban rail transit. It can prevent and reduce tunnel structural defects and extend the service life of tunnels. At present, tunnel deformation detection is mainly based on the spatial coordinate information and cross-sectional information of three-dimensional laser scanning point cloud data, and the circumferential deformation of the tunnel is obtained by fitting and calculating the cross-sectional information. However, in the existing tunnel deformation detection, in order to reduce the amount of calculation, it is easy to ignore the meaningful point cloud data, resulting in detailed deformations such as cracks, protrusions, pits, etc., which cannot be detected.

[0003] Therefore, the present application provides a method and system for automatic detection of tunnel deformation to solve the above problems. Summary of the invention

[0004] The purpose of this application is to provide a method and system for automatic detection of tunnel deformation, so as to solve the problem that the current tunnel deformation detection tends to ignore meaningful feature points and detailed deformation in order to reduce the amount of calculation. By filtering the sliced ​​point cloud data, the amount of calculation is reduced while retaining the meaningful feature points, reflecting the overall deformation and detailed deformation of the tunnel.

[0005] The present application first provides an automatic detection method for tunnel deformation, including: S1, obtaining tunnel point cloud data of multiple intervals and preprocessing; S2, registering and splicing the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; S3, slicing the complete tunnel point cloud data to obtain multiple slice point cloud data; S4, filtering the slice point cloud data based on an improved filtering algorithm; S5, fitting the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; S6, analyzing the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data; Among them, S4, filtering the slice point cloud data based on the improved filtering algorithm; including: S410, feature screening the slice point cloud data to obtain feature point data and non-feature point data; S420, sparse processing the non-feature point data to obtain sparse point data; S430, merging the feature point data and the sparse point data to form the filtered slice point cloud data.

[0006] In a possible implementation, S410, feature screening is performed on the slice point cloud data to obtain feature point data and non-feature point data; including: S411, calculating the normal vector of each point in the slice point cloud data based on the principal component analysis method; S412, calculating the curvature feature of each point based on the normal vector of each point; S413, comparing the curvature feature of each point with the set curvature threshold, and screening to obtain feature point data and non-feature point data.

[0007] In a possible implementation, S420, sparse processing is performed on non-feature point data to obtain sparse point data; including: S421, constructing a three-dimensional voxel grid according to the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes; S422, calculating the center of gravity of the grid according to the coordinates of the non-feature points contained in each grid; S423, using the center of gravity of the grid to replace the non-feature points in the grid to obtain sparse point data.

[0008] In a possible implementation, S5, fitting processing is performed on the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; including: S51, selecting more than 3 data points from the sparse point data to form initial parameters, and constructing a circle model based on the initial parameters; S52, calculating the distance between the remaining points in the sparse point data and the circle model, and taking the points with a distance less than the distance threshold as inner points, and using them to correct the circle model; S53, traversing all remaining points, if the number of inner points is less than the number threshold, returning to S51, if the number of inner points is greater than or equal to the number threshold, the circle model is used as the overall fitting model of the slice point cloud data; S54, performing spline curve fitting based on the feature point data to obtain a detail fitting model; S55, superimposing the detail fitting model with the overall fitting model to obtain a fitting section containing detail features and overall features.

[0009] In a possible implementation, calculating the distance between the remaining points in the sparse point data and the circle model includes: calculating the difference between the distance from the point to the center of the circle model and the radius as the distance between the point and the circle model.

[0010] The present application also provides an automatic detection system for tunnel deformation, including: a data acquisition unit, used to acquire tunnel point cloud data of multiple intervals and perform preprocessing; a data splicing unit, used to register and splice the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; a data slicing unit, used to slice the complete tunnel point cloud data to obtain multiple slice point cloud data; a data filtering unit, used to filter the slice point cloud data based on an improved filtering algorithm; a data fitting unit, used to fit the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; a deformation analysis unit, used to analyze the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data; Among them, the data filtering unit includes: a feature point extraction unit, which is used to perform feature screening on the slice point cloud data to obtain feature point data and non-feature point data; a non-feature point refinement unit, which is used to perform sparse processing on the non-feature point data to obtain sparse point data; a feature point merging unit, which is used to merge the feature point data and the sparse point data to form the slice point cloud data after filtering.

[0011] In a possible implementation, the feature point extraction unit includes: a normal vector analysis unit, used to calculate the normal vector of each point in the slice point cloud data based on the principal component analysis method; a curvature analysis unit, used to calculate the curvature feature of each point based on the normal vector of each point; and a feature point classification unit, used to compare the curvature feature of each point with a set curvature threshold to screen out feature point data and non-feature point data.

[0012] In one possible implementation, the non-feature point refinement element includes: a grid construction unit, which is used to construct a three-dimensional voxel grid based on the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes; a centroid analysis unit, which is used to calculate the centroid of the grid based on the coordinates of the non-feature points contained in each grid; and a sparse refinement element, which is used to replace the non-feature points in the grid with the grid centroid to obtain sparse point data.

[0013] In one possible implementation, the data fitting unit includes: an overall model construction unit, which is used to select more than three data points from the sparse point data to form initial parameters, and construct a circular model based on the initial parameters; an overall model correction unit, which is used to calculate the distance between the remaining points in the sparse point data and the circular model, and use the points with a distance less than a distance threshold as internal points to correct the circular model; an overall model output unit, which is used to traverse all remaining points, and if the number of internal points is less than the number threshold, return to the overall model construction unit; if the number of internal points is greater than or equal to the number threshold, use the circular model as the overall fitting model of the sliced ​​point cloud data; a detail model fitting unit, which is used to perform spline curve fitting based on feature point data to obtain a detail fitting model; and a section fitting unit, which is used to superimpose the detail fitting model with the overall fitting model to obtain a fitting section containing detail features and overall features.

[0014] In a possible implementation manner, the overall model correction unit is specifically used to calculate the difference between the distance from the point to the center of the circular model and the radius as the distance between the point and the circular model.

[0015] Compared with the prior art, the present application has the following beneficial effects: the tunnel deformation automatic detection method and system provided by the present application filters the slice point cloud data by setting an improved filtering algorithm, retains the detail deformation point cloud data, and simplifies the overall deformation point cloud data, thereby reducing the amount of data while retaining the detail points, greatly reducing the amount of calculation, and ensuring the accuracy of detail and overall deformation detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings: Figure 1 A flow chart of a method for automatically detecting tunnel deformation provided in an embodiment of the present application; Figure 2 This is a structural diagram of the automatic tunnel deformation detection system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] Hereinafter, the term "include" or "may include" used in various embodiments of the present application indicates the presence of the function, operation or element applied for, and does not limit the increase of one or more functions, operations or elements. In addition, as used in various embodiments of the present application, the terms "include", "have" and their cognates are intended only to indicate specific features, numbers, steps, operations, elements, components or a combination of the foregoing items, and should not be understood as first excluding the presence of one or more other features, numbers, steps, operations, elements, components or a combination of the foregoing items or the possibility of increasing one or more features, numbers, steps, operations, elements, components or a combination of the foregoing items.

[0018] In various embodiments of the present application, the expression "or" or "at least one of B or / and C" includes any combination or all combinations of the words listed at the same time. For example, the expression "B or C" or "at least one of B or / and C" may include B, may include C, or may include both B and C.

[0019] The terms used in the various embodiments of the present application are only used for the purpose of describing specific embodiments and are not intended to limit the various embodiments of the present application. As used herein, the singular form is intended to also include the plural form, unless the context clearly indicates otherwise. Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the meanings commonly understood by ordinary technicians in the field to which the various embodiments of the present application belong. Terms (such as terms defined in generally used dictionaries) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning, unless clearly defined in the various embodiments of the present application.

[0020] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with examples and drawings. The illustrative implementation scheme of the present application and its description are only used to explain the present application and are not intended to limit the present application.

[0021] See also Figure 1 As shown, Figure 1 Flow chart of the automatic detection method of tunnel deformation provided by the embodiment of the present application. The method includes: S1, obtaining tunnel point cloud data of multiple intervals and preprocessing; S2, registering and splicing the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; S3, slicing the complete tunnel point cloud data to obtain multiple slice point cloud data; S4, filtering the slice point cloud data based on the improved filtering algorithm; S5, fitting the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; S6, analyzing the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data; Among them, S4, filtering the slice point cloud data based on the improved filtering algorithm; including: S410, feature screening the slice point cloud data to obtain feature point data and non-feature point data; S420, sparse processing the non-feature point data to obtain sparse point data; S430, merging the feature point data and the sparse point data to form the filtered slice point cloud data.

[0022] Specifically, in the automatic detection method for tunnel deformation, first, step S1 obtains point cloud data of different sections of the tunnel through a three-dimensional laser scanner, and performs preprocessing to eliminate noise and outliers to ensure the accuracy and reliability of the data; then, step S2 registers and splices the point cloud data of different sections through registration technology to form complete tunnel point cloud data for comprehensive analysis; then, step S3 slices the complete tunnel point cloud data at a certain distance along the forward direction of the tunnel to form multiple slice point cloud data for subsequent deformation analysis; step S4 uses an improved filtering algorithm to refine these slice point cloud data to retain detailed deformation (such as cracks, pits or protrusions) and overall deformation (tunnel circumferential deformation), providing a basis for subsequent complete tunnel deformation analysis; then, step S5 fits the filtered slice point cloud data to obtain fitting sections that can accurately reflect the deformation of the tunnel section, and these fitting sections contain both the overall deformation characteristics of the tunnel and the detailed deformation characteristics; finally, step S6 comprehensively analyzes all fitting sections to comprehensively evaluate the overall and detailed deformation of the tunnel, which is helpful for the safe maintenance of the tunnel.

[0023] The improvement of the present application lies in that, in view of the fact that there are relatively few point cloud data on tunnel detail deformation and relatively more redundant point cloud data on overall deformation, the present application sets up an improved filtering algorithm to filter the sliced ​​point cloud data, retain the detail deformation point cloud data, and streamline the overall deformation point cloud data, thereby reducing the amount of data while retaining the detail points, thereby ensuring the accuracy of detail and overall deformation detection.

[0024] S1. Obtain tunnel point cloud data of multiple sections and perform preprocessing; specifically, use a three-dimensional laser scanner to scan the tunnel in sections, obtain tunnel point cloud data of multiple sections, and perform denoising preprocessing on the tunnel point cloud data to eliminate interference noise and ensure the accuracy and reliability of the data.

[0025] S2. The tunnel point cloud data of multiple sections after preprocessing are aligned and spliced ​​to obtain complete tunnel point cloud data. Specifically, since the 3D laser scanner cannot measure the entire environment at one time, the point cloud data of multiple sections need to be combined into complete tunnel point cloud data through point cloud splicing technology after segmented scanning. When splicing, the point cloud data at different positions also need to be transformed into a unified coordinate system through rotation and translation to form aligned and spliced ​​complete tunnel point cloud data.

[0026] S3. Slice the complete tunnel point cloud data to obtain a plurality of slice point cloud data. Specifically, first adjust the direction of the complete tunnel point cloud data so that the tunnel extends along the X-axis, and then slice the complete tunnel point cloud data at regular intervals in a direction perpendicular to the X-axis direction to divide the complete tunnel point cloud data into a plurality of slice point cloud data.

[0027] S4. Filter the slice point cloud data based on the improved filtering algorithm; specifically, in step S410, the feature point data with relatively significant geometric changes and the non-feature point data with relatively gentle geometric changes are identified and distinguished through the feature screening algorithm. These feature point data often represent structures with significant geometric changes on the tunnel wall, such as cracks, pits or protrusions on the tunnel wall, while these non-feature data points can represent structures with gentle geometric changes on the tunnel wall, such as circular or elliptical tunnel walls; then, in step S420, the non-feature point data is sparsely processed to reduce redundant information, reduce the computational burden, and retain sufficient data for deformation analysis; finally, in step S430, the filtered feature point data is merged with the non-feature point data after sparse processing to form the slice point cloud data after filtering. The slice point cloud data after filtering contains both the detailed features of the tunnel and simplifies the redundant data, providing an accurate and streamlined data basis for the subsequent tunnel deformation analysis.

[0028] In a possible implementation, S410 performs feature screening on the slice point cloud data to obtain feature point data and non-feature point data; including: S411, calculating the normal vector of each point in the slice point cloud data based on the principal component analysis method; S412, calculating the curvature feature of each point based on the normal vector of each point; S413, comparing the curvature feature of each point with the set curvature threshold, and screening to obtain feature point data and non-feature point data.

[0029] Specifically, in step S411, the principal component analysis (PCA) method is used to analyze the local neighborhood of each point and the normal vector of each point is calculated; then, in step S412, the local area of ​​each point is plane-fitted and the curvature feature of each point is calculated using the normal vector of each point; finally, in step S413, the curvature feature of each point is compared with the pre-set curvature threshold to screen out feature points whose curvature features are greater than the curvature threshold. These points usually correspond to significant change areas on the tunnel surface, such as cracks, bulges or depressions, while feature points whose curvature features are less than or equal to the curvature threshold are classified as non-feature points. These points usually correspond to gently changing areas on the tunnel surface. The slice point cloud data is divided into feature point data and non-feature point data through curvature feature screening, providing support for subsequent data simplification and deformation analysis.

[0030] In a possible implementation, S420 performs sparse processing on non-feature point data to obtain sparse point data; including: S421, constructing a three-dimensional voxel grid based on the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes; S422, calculating the center of gravity of the grid based on the coordinates of the non-feature points contained in each grid; S423, using the center of gravity of the grid to replace the non-feature points in the grid to obtain sparse point data.

[0031] Specifically, first determine the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes, which are Xmin, Xmax, Ymin, Ymax, Zmin, and Zmax, respectively, and calculate the number of rows and columns of the voxel grid according to the set voxel grid size to construct a three-dimensional voxel grid. Each non-feature point is divided into the corresponding voxel grid according to its coordinates; then, based on the non-feature points contained in the voxel grid, calculate the mean value of its coordinates on the X, Y, and Z axes as the centroid of the voxel grid; finally, use the centroid of the grid to replace other feature points in the grid to obtain simplified sparse point data. By replacing the centroid, the amount of calculation is reduced while retaining sufficient geometric information.

[0032] In a possible implementation, S5, fitting processing is performed on the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; including: S51, selecting more than 3 data points from the sparse point data to form initial parameters, and constructing a circle model based on the initial parameters; S52, calculating the distance between the remaining points in the sparse point data and the circle model, and taking the points with a distance less than the distance threshold as inner points, and using them to correct the circle model; S53, traversing all remaining points, if the number of inner points is less than the number threshold, returning to S51, if the number of inner points is greater than or equal to the number threshold, the circle model is used as the overall fitting model of the slice point cloud data; S54, performing spline curve fitting based on the feature point data to obtain a detail fitting model; S55, superimposing the detail fitting model with the overall fitting model to obtain a fitting section containing detail features and overall features.

[0033] Further, calculating the distance between the remaining points in the sparse point data and the circle model includes: calculating the difference between the distance from the point to the center of the circle model and the radius as the distance between the point and the circle model.

[0034] Specifically, in step S51, at least three data points located on the tunnel section are randomly selected from the sparse point data in the sliced ​​point cloud data as parameters for constructing the initial circle model; in step S52, for the remaining points in the sparse point data, the difference between the distance from the point to the center of the current circle model and the radius is calculated as the distance between the point and the current circle model, and the points with a distance less than the distance threshold are identified as inliers. The inliers can be regarded as points consistent with the circle model, and the parameters of the circle model are updated using the inliers so that the sum of the deviations from all inliers to the circle model is minimized, and the updated circle model can better fit these inliers; in step S53, after traversing all the remaining points, the number of inliers is evaluated. If the number of inliers is less than the set value, the inliers are updated. If the number threshold is exceeded, it means that the current circle model cannot fit all the data well, and it is necessary to return to step S51 to reselect the initial parameters and build a new circle model. Otherwise, it means that the current circle model can fit all the data well, and the circle model can be used as the overall fitting model of the slice point cloud data to represent the overall deformation of the tunnel. In step S54, for the feature point data in the slice point cloud data, spline curve fitting technology (such as B-spline or R-spline) is used to perform detail fitting, and a curve is generated as a detail fitting model to represent the detail deformation of the tunnel. Finally, in step S55, the detail fitting model is superimposed on the overall fitting model to form a fitting section containing detail features and overall features.

[0035] S6. Analyze the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data. Specifically, according to the position of the fitting section in the tunnel, place the center of the fitting section at the corresponding position of the tunnel design centerline. According to the radius of the fitting section and the design radius of the position, obtain the circumferential deformation of the tunnel at the position. According to the detailed features on the fitting section, analyze the possible cracks, protrusions or depressions at the position.

[0036] It can be understood that the automatic tunnel deformation detection method provided by the present application extracts and retains the point cloud data representing the detailed deformation, simplifies the point cloud data representing the overall deformation, and ensures the accuracy of the detailed and overall deformation detection while reducing the amount of data.

[0037] See also Figure 2 As shown, Figure 2 This is a structural diagram of the automatic tunnel deformation detection system provided in the embodiment of the present application. The system includes: a data acquisition unit, which is used to acquire tunnel point cloud data of multiple intervals and perform preprocessing; a data splicing unit, which is used to align and splice the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; a data slicing unit, which is used to slice the complete tunnel point cloud data to obtain multiple slice point cloud data; a data filtering unit, which is used to filter the slice point cloud data based on an improved filtering algorithm; a data fitting unit, which is used to fit the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; a deformation analysis unit, which is used to analyze the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data; Among them, the data filtering unit includes: a feature point extraction unit, which is used to perform feature screening on the slice point cloud data to obtain feature point data and non-feature point data; a non-feature point refinement unit, which is used to perform sparse processing on the non-feature point data to obtain sparse point data; a feature point merging unit, which is used to merge the feature point data and the sparse point data to form the slice point cloud data after filtering.

[0038] In a possible implementation, the feature point extraction unit includes: a normal vector analysis unit, used to calculate the normal vector of each point in the slice point cloud data based on the principal component analysis method; a curvature analysis unit, used to calculate the curvature feature of each point based on the normal vector of each point; and a feature point classification unit, used to compare the curvature feature of each point with a set curvature threshold to screen out feature point data and non-feature point data.

[0039] In one possible implementation, the non-feature point refinement element includes: a grid construction unit, which is used to construct a three-dimensional voxel grid based on the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes; a centroid analysis unit, which is used to calculate the centroid of the grid based on the coordinates of the non-feature points contained in each grid; and a sparse refinement element, which is used to replace the non-feature points in the grid with the grid centroid to obtain sparse point data.

[0040] In one possible implementation, the data fitting unit includes: an overall model construction unit, which is used to select more than three data points from the sparse point data to form initial parameters, and construct a circular model based on the initial parameters; an overall model correction unit, which is used to calculate the distance between the remaining points in the sparse point data and the circular model, and use the points with a distance less than a distance threshold as internal points to correct the circular model; an overall model output unit, which is used to traverse all remaining points, and if the number of internal points is less than the number threshold, return to the overall model construction unit; if the number of internal points is greater than or equal to the number threshold, use the circular model as the overall fitting model of the sliced ​​point cloud data; a detail model fitting unit, which is used to perform spline curve fitting based on feature point data to obtain a detail fitting model; and a section fitting unit, which is used to superimpose the detail fitting model with the overall fitting model to obtain a fitting section containing detail features and overall features.

[0041] In a possible implementation manner, the overall model correction unit is specifically used to calculate the difference between the distance from the point to the center of the circular model and the radius as the distance between the point and the circular model.

[0042] It should be noted that the automatic tunnel deformation detection system provided in the present application is used to implement the above-mentioned automatic tunnel deformation detection method, corresponds one-to-one with the above-mentioned automatic tunnel deformation detection method, and has corresponding effects, so it will not be elaborated on here.

[0043] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for automatic detection of tunnel deformation, characterized in that: include: S1, obtaining tunnel point cloud data of multiple sections and performing preprocessing; S2, registering and splicing the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; S3, slicing the complete tunnel point cloud data to obtain a plurality of slice point cloud data; S4, filtering the slice point cloud data based on an improved filtering algorithm; S5, performing fitting processing on the filtered slice point cloud data to obtain a fitting section including detail features and overall features; S6. Analyze the overall and detailed deformation of the tunnel based on the fitting sections of multiple slice point cloud data; Among them, S4, filtering the slice point cloud data based on an improved filtering algorithm; including: S410, feature screening the slice point cloud data to obtain feature point data and non-feature point data; S420, sparse processing the non-feature point data to obtain sparse point data; S430, merging the feature point data and the sparse point data to form the filtered slice point cloud data.

2. The method for automatic detection of tunnel deformation according to claim 1, characterized in that: S410, feature screening is performed on the slice point cloud data to obtain feature point data and non-feature point data; including: S411, calculating the normal vector of each point in the slice point cloud data based on the principal component analysis method; S412, calculating the curvature feature of each point based on the normal vector of each point; S413: Compare the curvature feature of each point with the set curvature threshold, and screen out feature point data and non-feature point data.

3. The method for automatic detection of tunnel deformation according to claim 1, characterized in that: S420, performing sparse processing on non-feature point data to obtain sparse point data; including: S421, constructing a three-dimensional voxel grid according to the maximum and minimum values ​​of the non-feature point data on the X, Y, and Z axes; S422, calculating the center of gravity of each grid according to the coordinates of the non-feature points contained in each grid; S423, using the grid centroid to replace the non-feature points in the grid to obtain sparse point data.

4. The method for automatic detection of tunnel deformation according to claim 1, characterized in that: S5, performing fitting processing on the filtered slice point cloud data to obtain a fitting section containing detail features and overall features; including: S51, selecting three or more data points from the sparse point data to form initial parameters, and constructing a circle model based on the initial parameters; S52, calculating the distance between the remaining points in the sparse point data and the circle model, taking the points whose distance is less than the distance threshold as the inner points, and using them to correct the circle model; S53, traversing all remaining points, if the number of inner points is less than the number threshold, returning to S51, if the number of inner points is greater than or equal to the number threshold, taking the circle model as the overall fitting model of the slice point cloud data; S54, performing spline curve fitting based on the feature point data to obtain a detail fitting model; S55, superimposing the detail fitting model and the overall fitting model to obtain a fitting section including detail features and overall features.

5. The method for automatic detection of tunnel deformation according to claim 4, characterized in that: The calculating the distance between the remaining points in the sparse point data and the circle model includes: calculating the difference between the distance from the point to the center of the circle model and the radius as the distance between the point and the circle model.

6. An automatic tunnel deformation detection system, characterized in that: include: A data acquisition unit, used for acquiring tunnel point cloud data of multiple sections and performing preprocessing; A data splicing unit, used for registering and splicing the tunnel point cloud data of multiple intervals after preprocessing to obtain complete tunnel point cloud data; A data slicing unit, used for slicing the complete tunnel point cloud data to obtain a plurality of slice point cloud data; A data filtering unit, used for filtering the slice point cloud data based on an improved filtering algorithm; A data fitting unit, used for fitting the slice point cloud data after filtering to obtain a fitting section containing detail features and overall features; Deformation analysis unit, used to analyze the overall and detailed deformation of the tunnel based on the fitting section of multiple slice point cloud data; Among them, the data filtering unit includes: a feature point extraction unit, which is used to perform feature screening on the slice point cloud data to obtain feature point data and non-feature point data; a non-feature point refinement unit, which is used to perform sparse processing on the non-feature point data to obtain sparse point data; a feature point merging unit, which is used to merge the feature point data and the sparse point data to form the slice point cloud data after filtering.

7. The automatic tunnel deformation detection system according to claim 6, characterized in that: The feature point extraction unit comprises: A normal vector analysis unit, used for calculating the normal vector of each point in the slice point cloud data based on the principal component analysis method; A curvature analysis unit, for calculating a curvature feature of each point based on a normal vector of each point; The feature point classification unit is used to compare the curvature feature of each point with a set curvature threshold to screen out feature point data and non-feature point data.

8. The automatic tunnel deformation detection system according to claim 6, characterized in that: The non-feature point refined element includes: A grid construction unit, used to construct a three-dimensional voxel grid according to the maximum and minimum values ​​of non-feature point data on the X, Y, and Z axes; A centroid analysis unit, used for calculating the centroid of each grid according to the coordinates of the non-feature points contained in each grid; The sparse refined element is used to replace the non-feature points in the grid with the grid centroid to obtain sparse point data.

9. The automatic tunnel deformation detection system according to claim 6, characterized in that: The data fitting unit comprises: An overall model building unit, used for selecting more than 3 data points from the sparse point data to form initial parameters, and building a circle model based on the initial parameters; The overall model correction unit is used to calculate the distance between the remaining points in the sparse point data and the circle model, and take the points whose distance is less than the distance threshold as the inner points, and use them to correct the circle model; The overall model output unit is used to traverse all the remaining points. If the number of internal points is less than the number threshold, the overall model construction unit is returned. If the number of internal points is greater than or equal to the number threshold, the circle model is used as the overall fitting model of the slice point cloud data; A detail model fitting unit, used for performing spline curve fitting based on feature point data to obtain a detail fitting model; The section fitting unit is used to superimpose the detail fitting model with the overall fitting model to obtain a fitting section including detail features and overall features.

10. The automatic tunnel deformation detection system according to claim 9, characterized in that: The overall model correction unit is specifically used to calculate the difference between the distance from the point to the center of the circle model and the radius as the distance between the point and the circle model.

Images