Roadway contour extraction method, system and equipment based on three-dimensional point cloud and medium

Abstract

The invention belongs to the technical field of data processing, and provides a roadway contour extraction method, system, equipment and medium based on three-dimensional point cloud in order to solve the problems of inaccuracy, low efficiency and the like in existing coal mine roadway feature extraction. Combined noise filtering is carried out on a space formed by connecting angular points of fitting contours on adjacent slices, so that scattered noise and useless shielding are effectively reduced, and the data definition and reliability are improved; based on neighborhood calculation perpendicularity features, a top plate, the ground and the side wall of the roadway are recognized, and non-roadway structure data such as pipelines are accurately removed in combination with a cylinder fitting method and a plane fitting method, so that the accuracy of roadway contour extraction is ensured; and fitting the outermost plane based on the farthest points of the point clouds on the upper, lower, left and right surfaces by adopting a segmentation-projection method, and extracting irregular roadway contours from the top surface, the ground and the side surfaces to realize refined roadway contour extraction.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of data processing, and particularly relates to a roadway profile extraction method, system, device and medium based on three-dimensional point clouds. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] As an important part of mine safety, the structural stability of the coal mine roadway directly affects the safety production of the mine. Among them, the construction quality of the surrounding rock and the supporting structure of the roadway determines the long-term stability of the roadway. However, the coal mine roadway engineering has the characteristics of complex excavation environment, variable surrounding rock conditions, limited construction space, etc., so there are the following deficiencies in the detection of the coal mine roadway: the mechanical and automatic level of the detection equipment is low, relying on manual inspection, the detection efficiency is low and the labor cost is high; the conventional detection means is mostly sampling detection, which is difficult to realize full coverage monitoring, resulting in low representativeness and reliability of the detection data; in addition, due to the limitation of the underground environment, the application of part of the detection equipment in the roadway is limited, and there are great difficulties in long-term monitoring and later review and calibration.

[0004] In view of the above deficiencies, based on the characteristics of three-dimensional laser point cloud technology, the adaptability of three-dimensional laser point cloud technology to coal mine roadway is considered, which can realize large range, high precision, full range and non-contact acquisition of roadway surface point cloud data, so as to obtain three-dimensional point cloud information of the surrounding rock and the supporting structure of the roadway. However, there are still the following deficiencies in accurately extracting the key structural information of the roadway from the point cloud data: the roadway scene is complex, and various structures are mutually occluded, causing part of the point cloud data to be missing, so that it is difficult to uniformly extract complete supporting information; the amount of roadway point cloud data is huge, and the point cloud of the supporting structure only accounts for a part of it, resulting in large amount of calculation of the algorithm for extracting target data and low processing efficiency; in addition, the roadway environment is complex, and the point cloud may contain noise interference such as equipment and floating dust, further increasing the difficulty of data cleaning and accurate analysis. The existing feature clustering analysis algorithm applied in the field of feature recognition of coal mine roadway has various problems, and cannot achieve the purpose of high-precision extraction of roadway point cloud. SUMMARY

[0005] In order to overcome the deficiencies of the prior art, the present application provides a roadway profile extraction method, system, device and medium based on three-dimensional point clouds, which overcomes the deficiencies of the existing coal mine roadway point cloud processing method in noise filtering, structure extraction and profile extraction, and provides efficient and accurate technical support for the subsequent digital management, deformation monitoring and intelligent construction of the coal mine roadway.

[0006] In order to achieve the above purpose, the present application adopts the following technical solutions: In a first aspect, the present application provides a coal mine roadway feature extraction method based on a three-dimensional point cloud, comprising: Obtaining roadway point cloud original data, filtering out noise points based on the local density of the roadway point cloud original data, and a space body formed by connecting corner points of a fitted contour on adjacent slices; Correcting the roadway point cloud data after noise point filtering, calculating the perpendicularity based on the roadway neighborhood radius, extracting the roadway ground, roof and wall surface according to the calculated perpendicularity, and removing the pipeline point cloud data by combining cylindrical fitting and plane fitting to obtain a coarse contour of the roadway; Segmenting the coarse contour of the roadway, fitting the outermost plane based on the farthest points of the point clouds on the upper, lower, left and right four sides of the roadway, and extracting a fine contour of the roadway by combining the reference plane gridding projection technology.

[0007] In a second aspect, the present application provides a coal mine roadway feature extraction system based on a three-dimensional point cloud, comprising: A noise point filtering module configured to obtain roadway point cloud original data, filter out noise points based on the local density of the roadway point cloud original data, and a space body formed by connecting corner points of a fitted contour on adjacent slices; A coarse contour of the roadway module configured to correct the roadway point cloud data after noise point filtering, calculate the perpendicularity based on the roadway neighborhood radius, extract the roadway ground, roof and wall surface according to the calculated perpendicularity, and remove the pipeline point cloud data by combining cylindrical fitting and plane fitting to obtain a coarse contour of the roadway; A fine contour of the roadway module configured to segment the coarse contour of the roadway, fit the outermost plane based on the farthest points of the point clouds on the upper, lower, left and right four sides of the roadway, and extract a fine contour of the roadway by combining the reference plane gridding projection technology.

[0008] In a third aspect, the present application provides an electronic device comprising a memory and a processor, and computer instructions stored on the memory and running on the processor, wherein when the computer instructions are run by the processor, the method of the first aspect is completed.

[0009] In a fourth aspect, the present application provides a computer readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method of the first aspect is completed.

[0010] The above one or more technical solutions have the following beneficial effects: In the present application, the scattered noise and useless occlusion are effectively reduced, the data clarity and reliability are improved by combined noise filtering through local density and the space formed by connecting the corner points of the fitting contour on adjacent slices; the roof, ground and sidewall of the roadway are identified based on the neighborhood calculation of the perpendicularity feature, and the non-roadway structure data such as pipeline is accurately removed by combining the cylindrical fitting and plane fitting method, so as to ensure the accuracy of the roadway contour extraction; the irregular roadway contour is extracted from the top surface, ground surface and side surface by using the subsection-projection method based on the farthest point fitting of the outermost plane of the upper, lower, left and right four points, so as to realize the fine roadway contour extraction. The present application overcomes the deficiencies of the existing coal mine roadway point cloud processing method in noise filtering, structure extraction and contour extraction, and provides efficient and accurate technical support for the subsequent digital management, deformation monitoring and intelligent construction of the coal mine roadway.

[0011] Advantages of the additional aspects of the present application will be partially given in the following description, partially become obvious from the following description, or be known by the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0012] The drawings accompanying the specification of this application form a part thereof, serve to provide further understanding of the application, and together with the description of the exemplary embodiments of the application given below, make an explanation of the application, and do not constitute an improper limitation of the application.

[0013] Figure 1 The figure is the overall flow chart of the coal mine roadway contour extraction method based on three-dimensional point cloud in the embodiment one of the present application; Figure 2 The figure is the working flow chart of the noise filtering in the embodiment one of the present application; Figure 3 The figure is the feature point extraction and matching flow chart in the data rectification work in the embodiment one of the present application; Figure 4 The figure is the roadway contour rough extraction stage flow chart in the embodiment one of the present application; Figure 5 The figure is the roadway contour fine extraction stage flow chart in the embodiment one of the present application; Figure 6 The figure is the roadway contour fine extraction effect diagram in the embodiment one of the present application; Figure 7 The figure is the roadway contour fine extraction effect diagram in the embodiment one of the present application; Figure 8 The figure is the comparison diagram before and after the roadway contour fine extraction in the embodiment one of the present application. DETAILED DESCRIPTION

[0014] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0015] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application.

[0016] In the case of no conflict, the embodiments in the application and the features in the embodiments can be combined with each other.

[0017] As introduced in the background, the tunnel scene is complex, various structures are mutually occluded, part of the point cloud data is missing, and it is difficult to extract complete support information uniformly; the tunnel point cloud data is huge, and the support structure point cloud only accounts for a part of it, resulting in large amount of calculation of the algorithm for extracting target data and low processing efficiency; in addition, the tunnel environment is complex, and the point cloud may contain noise interference such as equipment and floating dust, which further increases the difficulty of data cleaning and accurate analysis. In order to solve the problems of inaccurate tunnel feature extraction and low efficiency of the existing scheme, the embodiment provides a tunnel contour extraction method, system, device and medium based on three-dimensional point cloud.

[0018] The technical solutions in the embodiments of the application will be described clearly and completely in conjunction with the drawings of the embodiments of the application.

[0019] Embodiment one Figure 1 It is a whole flow chart of a tunnel contour extraction method based on three-dimensional point cloud provided by the embodiment, and the method comprises: Step 1: obtaining tunnel point cloud original data, based on the local density of the tunnel point cloud original data, and the space body formed by connecting the corner points of the fitted contour on the adjacent slices to remove noise points.

[0020] As shown in Figure 2 In one embodiment, the process of noise point filtering comprises: Firstly, the point cloud data is voxelized as a standard body composed of squares, and then a first feature threshold M is set, the squares with the number of points less than the first feature threshold M are deleted, and all the remaining squares are merged into a first point cloud set; wherein the size of the square is set to 5cm in this embodiment, and can be set accordingly according to the actual situation.

[0021] Secondly, all points in the first point cloud set are traversed, the local density of k-NN neighborhood of all points is calculated, and a second feature threshold M1 is set, the points with local density lower than the second feature threshold M1 are regarded as outliers, and the second point cloud set, i.e. point cloud 2, is obtained after deleting the outliers. Then, the second point cluster is uniformly sliced ​​at a set distance (e.g., 1m). Multiple fitted lines are obtained on each slice using a linear fitting method. These fitted lines form a first fitted contour, i.e., fitted contour 1. The first fitted contour is scaled to obtain a second fitted contour, ensuring the minimum distance between the second and first fitted contours is a preset value, such as 0.8m. Adjacent second fitted contours are connected by corner points to form an irregular spatial volume. All points within the irregular spatial volume are then deleted.

[0022] It should be noted that there are many auxiliary structures on the wall surface in coal mine roadways, such as pipes. Ensuring that there is a minimum distance between the second fitted profile and the first fitted profile is to avoid segmentation into pipes.

[0023] Step 2: Data correction, used to extract feature points and match them with absolute coordinates to realize the transformation of coal mine roadway point cloud coordinates to geodetic coordinates.

[0024] like Figure 3 As shown, in one embodiment, the data correction process includes point measurement and feature point extraction and matching.

[0025] The data correction measurement involves setting up prism target points and rectangular special target plates along the longitudinal direction of the tunnel, based on the baseline at the construction site. The material of the target plate is determined through experiments to ensure that its reflection intensity is within a specific range F.

[0026] Using a total station, at least two prism target points are measured, and the coordinates of the erection points are calculated in reverse. Then, the corner points of the rectangular special target plate are measured to obtain the absolute position information.

[0027] Feature point extraction and matching for data correction: Point cloud data is filtered using the reflection intensity range F. The filtered results are then subjected to Euclidean distance clustering analysis based on a set feature threshold (e.g., 5cm). According to the clustering analysis results, boundary lines are extracted for each type of point cloud data, and four corner points are extracted. These four corner points are then traversed, and the five points furthest from each corner point are selected. If any of these five points contain the corresponding diagonal corner point, the feature point extraction of the rectangular target plate is successful. The extracted corner point coordinates of the rectangular target plate are relative coordinates. These are matched with the corresponding total station measurements to calculate the spatial coordinate transformation matrix, converting the relative coordinate information of the tunnel into absolute coordinate information.

[0028] The embodiment utilizes a total station to measure at least two prism target points, and reversely calculates the coordinate information of the erected point. In the definition of surveying, it is called resection. In the process of building a roadway, a guide point, i.e., a base point, is laid out, which is an absolute coordinate. The relative spatial position relationship between the prism center point and the base point can be calculated by installing a specific size accessory, and then the absolute coordinate of the prism center point is obtained. The absolute coordinate system and the absolute coordinate of the erected point can be determined by at least two prisms in a resection manner. At this time, the rectangular special target plate corner point coordinates measured by the total station are absolute coordinates in the absolute coordinate system.

[0029] By screening at least 4 pairs of rectangular special target plate corner point coordinates which are not coplanar, a high-precision spatial coordinate system conversion parameter is solved by using a nonlinear solving method of the least square method, and then the point cloud data coordinates are converted to the absolute coordinate system, the data is corrected, and the corrected roadway point cloud is obtained.

[0030] Step 3: Calculate the perpendicularity based on the neighborhood radius of the roadway, extract the roadway ground, roof and wall surface according to the calculated perpendicularity, and remove the pipeline point cloud data by combining cylindrical fitting and plane fitting to obtain the rough contour of the roadway.

[0031] As shown in Figure 4 , all points are traversed, the neighborhood radius r is set based on the size of the roadway, the covariance matrix C of the roadway is calculated in the neighborhood of each point, then the eigenvalue decomposition of the covariance matrix C of the roadway is performed to obtain the eigenvectors from large to small: λ1, λ2, λ3 λ1, λ2, λ3 Three eigenvalues. Among them, the eigenvector corresponding to the maximum eigenvalue: main direction (plane or linear direction); the eigenvector corresponding to the minimum eigenvalue: normal vector (n).

[0032] The perpendicularity is calculated based on the set reference direction v=(0,0,1) and the normal vector. The ground point cloud is accurately extracted by setting the interval C1, C2, C3 and C4 of the perpendicularity, and the point cloud of the roof and the two side walls is roughly extracted.

[0033] Among them, the values of C1, C2, C3 and C4 can be appropriately selected according to the specific application scene, as well as the data quality and the neighborhood radius.

[0034] Among them, the angle between the normal vector and the reference direction is calculated :

[0035] Perpendicularity calculation: D=90°-

[0036] In addition, the pipeline in the roadway is thick, which is easy to be misidentified as a wall, but the diameter of the pipeline is known, and the pipeline data can be separated based on the RANSAC cylindrical fitting, plane fitting method and combined with the Euclidean distance clustering method.

[0037] Specifically, based on the two-side wall point cloud data obtained by the above rough extraction, first, the RANSAC cylindrical fitting is performed according to the diameter of the pipeline to find the pipeline point cloud, then the remaining wall point cloud is fitted based on the plane fitting method to find the wall, and the distance from the wall to the vertical center line of the pipeline is taken as the distance threshold D for classification to extract the wall data and the pipeline data.

[0038] In the three-dimensional point cloud data, due to the occlusion of the pipeline, only half of the point cloud data of the outside of the pipeline can be collected, so the distance from the wall to the vertical center line of the pipeline is taken as the distance threshold D to extract the wall data and the pipeline data.

[0039] In addition, there are anchor rods, wires and other noise points in the roadway, which are closely connected with the wall, and the three eigenvalues can be used for linear calculation to separate and direct the roadway structure with less noise points. Figure 7

[0040] The linearity L of the point is calculated by the following formula, and the larger the L is, the stronger the linearity of the point is.

[0041]

[0042] The covariance matrix of the anchor rod is solved based on the radius of the anchor rod and the neighborhood radius , and the covariance matrix of the wire is solved based on the radius of the wire and the neighborhood radius . The linearity values of the points are obtained in the same way as described above, and the linearity values of the points are compared with the set linearity threshold value corresponding to the anchor rod or the set linearity threshold value corresponding to the wire . If > , or > , it is a linear point cloud, otherwise it is other point cloud. The linear point cloud is extracted by traversing the point cloud.

[0043] Step 4: The rough contour of the roadway is segmented and projected, the outermost plane is fitted based on the farthest points of the upper, lower, left and right point clouds of the roadway, and the fine contour of the upper, lower, left and right of the roadway is extracted by combining the grid projection technology based on the reference plane. After extracting the fine contour, the abnormal points are automatically corrected by combining the point cloud reconstruction result and the fitting error statistics.

[0044] As shown in Figure 5 , the outermost data of the three-dimensional contour of the roadway is similar to a square box wrapped on the surface of the roadway data, and the points closest to the box in the same section of the roadway are extracted.​

[0045] Specifically, such as Figure 6 As shown, the point clouds of the ground, roof, and side walls extracted in step 3 are combined to form a new tunnel data. Then, the tunnel is evenly segmented and sliced ​​along its direction. The ground point cloud is used as a reference to fit a plane, and the plane is moved down by 1m to obtain the reference horizontal plane 1. The grid is divided according to the average density of the point cloud. Then, the top and ground point clouds are projected onto the reference horizontal plane 1 along the plane normal. The normal distance from these two planes to the projection plane is calculated. The grid node positions of the reference horizontal plane 1 are selected to find the point with the maximum distance between the top surface and the reference horizontal plane 1 and the point with the minimum distance between the ground and the reference horizontal plane 1, so as to obtain the fine contours of the two irregular planes.

[0046] Similarly, based on the segmented data, after fitting a plane with the point cloud of the left wall as the reference, the plane is shifted 1m to the left to obtain the reference elevation 1. After dividing the grid according to the average density of the point cloud, the walls on the left and right sides are projected along the normal of the reference elevation 1. The grid node positions of the reference elevation 1 are selected, with the right wall having the maximum distance from the reference elevation 1 and the left wall having the minimum distance from the reference elevation 1, to obtain the fine contours of the two irregular planes on the left and right.

[0047] Figure 8 This is a diagram showing the refined extraction effect of the tunnel contour in this embodiment. ​ This example presents a comparison image before and after the fine extraction of the tunnel contour in this embodiment. This embodiment employs a segmented projection method, fitting the outermost plane based on the farthest point of the point cloud on all four sides of the tunnel, and combining this with a reference plane mesh projection technique. Furthermore, a confidence weight based on point reflection intensity is introduced during the projection process to improve the robustness of the tunnel contour projection. This allows for the extraction of fine contours in all directions, effectively reducing LiDAR measurement errors and improving the accuracy of digital tunnel modeling.

[0048] This operation can reduce the accuracy error of handheld LiDAR and reduce the point cloud thickness error by at least 2cm.

[0049] Example 2 This embodiment also provides a tunnel contour extraction system based on 3D point clouds, including: The noise removal module is configured to: acquire the original data of the tunnel point cloud, and perform noise removal based on the local density corresponding to the original data of the tunnel point cloud and the spatial volume formed by connecting the corner points of the fitted contour on adjacent slices. The tunnel rough outline module is configured to: correct the noise-filtered tunnel point cloud data, calculate the verticality based on the tunnel neighborhood radius, extract the tunnel floor, roof and wall based on the calculated verticality, and remove the pipe point cloud data by combining cylindrical fitting and planar fitting to obtain the tunnel rough outline. The roadway fine profile module is configured to segment the roadway coarse profile, fit the outermost plane based on the farthest points of the upper, lower, left and right four surface point clouds of the roadway, and extract the roadway fine profile by combining the reference plane and the grid projection technology.

[0050] In more embodiments, there are also provided: An electronic device includes a memory and a processor, and computer instructions stored in the memory and run on the processor, when the computer instructions are run by the processor, the method described in embodiment one is completed. For brevity, this will not be described here.

[0051] It should be understood that in the embodiments, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, ready-to-program gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0052] The memory can include read-only memory and random access memory, and provide instructions and data to the processor, and a portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0053] A computer readable storage medium for storing computer instructions, when the computer instructions are executed by a processor, the method described in embodiment one is completed.

[0054] The method in embodiment one can be directly embodied as a hardware processor to complete, or be completed by a combination of hardware and software modules in the processor. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is in the memory, and the processor reads the information in the memory, and combines the hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0055] A computer program product includes a computer program, when the computer program is executed by a processor, the method described in embodiment one is implemented.

[0056] The present application also provides at least one computer program product tangibly stored on a non-transitory computer readable storage medium. The computer program product includes computer executable instructions, for example, instructions embodied in program modules, executed by devices at the target real or virtual processor to perform the processes / methods as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The functionality of the program modules can be combined or split between program modules as desired in various embodiments. Machine executable instructions for program modules can be executed within a local or distributed device. In a distributed device, program modules can be located in local and remote memory storage devices.

[0057] Computer program code for carrying out operations of the present application can be written in one or more programming languages. These computer program codes can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program codes, when executed by the computer or other programmable data processing apparatus, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program codes can be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0058] In the context of the present application, the computer program code or related data can be carried by any suitable carrier to enable the device, apparatus or processor to perform the various processes and operations described above. Examples of the carrier include a signal, a computer readable medium, etc. Examples of the signal can include an electrical, optical, radio, sound or other forms of propagated signals, such as a carrier wave, an infrared signal, etc.

[0059] Those skilled in the art can realize that the units and algorithm steps of the examples described in conjunction with the present embodiments can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0060] The above describes the specific embodiments of the present application in conjunction with the accompanying drawings, but is not a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without inventive labor are still within the scope of protection of the present application.

Claims

1. A method for extracting features from coal mine roadways based on 3D point clouds, characterized in that, include: Obtain the original point cloud data of the tunnel, and filter out noise based on the local density of the original point cloud data and the spatial volume formed by connecting the corner points of the fitted contour on adjacent slices. The noise-filtered tunnel point cloud data is corrected, the verticality is calculated based on the tunnel neighborhood radius, the tunnel floor, roof and walls are extracted based on the calculated verticality, and the pipe point cloud data is removed by combining cylindrical fitting and planar fitting to obtain the rough outline of the tunnel. The rough contour of the tunnel is segmented, and the outermost plane is fitted based on the farthest point of the point cloud on the upper, lower, left and right sides of the tunnel. Combined with the reference plane gridding projection technology, the fine contour of the tunnel is extracted.

2. The method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 1, characterized in that, Noise removal is performed based on the local density of the original point cloud data of the alleyway, specifically as follows: The original point cloud data of the alleyway is divided into a standard volume composed of multiple squares; Delete the squares whose number of points in the grid is less than the first feature threshold to obtain the first point cloud set; Calculate the local density of the k-NN neighborhood of the first point cloud, delete outliers with a density less than the second feature threshold, and obtain the second point cloud.

3. The method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 2, characterized in that, Noise removal is performed on the spatial volume formed by connecting the corner points of the fitted contours on adjacent slices, specifically as follows: The second point cloud is divided into evenly spaced slices, and multiple fitted lines are obtained in each slice by straight line fitting to obtain the first fitted contour. The first fitted contour is scaled to obtain the second fitted contour; Connect the corner points of adjacent second fitted contours to obtain an irregular space volume, and then delete all point clouds within the irregular space volume.

4. The method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 1, characterized in that, Verticality is calculated based on the neighborhood radius of the tunnel. The tunnel floor, roof, and walls are then extracted based on the calculated verticality to obtain the rough tunnel profile, specifically: The covariance matrix of the tunnel is determined based on the neighborhood radius determined by the tunnel size. Based on the covariance matrix, three eigenvalues ​​are determined, and based on the three eigenvalues, the normal vector of the tunnel is determined. Calculate the verticality based on the normal vector of the tunnel and the set reference direction; Based on different preset verticality ranges, point clouds are extracted from the roadway floor, roof, and side walls to obtain the rough outline of the roadway.

5. A method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 1 or 4, characterized in that, By combining cylindrical fitting and planar fitting to remove pipe point cloud data, pipe-free wall point cloud data is obtained, specifically: RANSAC cylindrical fitting is performed based on the pipe diameter to find the pipe point cloud data; The remaining wall point cloud data is used to obtain the wall surface using a planar fitting method; The distance from the wall to the vertical centerline of the pipe is used as a distance threshold for classification, and the point cloud data of the wall and the pipe are extracted.

6. The method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 1, characterized in that, The noise-filtered point cloud data of the roadway is then corrected, specifically as follows: Prism target points and rectangular special target plates are laid out longitudinally along the tunnel construction site based on the base points; the material of the rectangular special target plate has a reflectance intensity within a specific range F. Using a total station, at least two prism target points are measured, and the coordinate information of the erection point is calculated in reverse. Then, the corner points of the rectangular special target plate are measured to obtain the absolute position information of the feature points of the rectangular special target plate. The feature points of the rectangular special target plate are matched with the corresponding total station measurements to obtain the coordinate transformation matrix, thereby converting the relative coordinate information of the tunnel point cloud data into absolute coordinate information.

7. The method for extracting coal mine roadway features based on three-dimensional point clouds as described in claim 6, characterized in that, The feature point extraction of the rectangular special target plate was confirmed to be correct, specifically as follows: The point cloud data was filtered by a specific range of reflection intensity to obtain the point cloud data of the rectangular special target plate; The screening results are then subjected to cluster analysis based on the set clustering feature values; Based on the clustering analysis results, the boundary line and four corner points are extracted for each type of point cloud. The four corner points are traversed and the five points with the farthest distance are selected with each corner point as the center. If the five selected points include the corresponding diagonal corner points, then the feature points of the rectangular special target board have been extracted correctly.

8. A coal mine roadway feature extraction system based on 3D point clouds, characterized in that, include: The noise removal module is configured to: acquire the original roadway point cloud data, and perform noise removal based on the local density of the original roadway point cloud data and the spatial volume formed by connecting the corner points of the fitted contours on adjacent slices. The tunnel rough outline module is configured to: correct the noise-filtered tunnel point cloud data, calculate the verticality based on the tunnel neighborhood radius, extract the tunnel floor, roof and wall based on the calculated verticality, and remove the pipe point cloud data by combining cylindrical fitting and planar fitting to obtain the tunnel rough outline. The fine contour module of the tunnel is configured to: segment the coarse contour of the tunnel, fit the outermost plane based on the farthest point of the point cloud on the upper, lower, left and right sides of the tunnel, and extract the fine contour of the tunnel by combining the reference plane grid projection technology.

9. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, perform the method described in any one of claims 1-7.

Images