Tunnel lining surface crack detection device, method, medium and equipment

Abstract

The present invention relates to the technical field of tunnel engineering monitoring, and specifically to a tunnel lining surface crack detection device, comprising a track, a detection module, a positioning module, a processing module and an identification module; the track is arranged on the top of the tunnel; the detection module is slidably arranged on the track, and is used to scan and photograph the tunnel lining surface to obtain a tunnel image and a three-dimensional point cloud respectively; the positioning module is arranged at the top central axis of the tunnel entrance, and the positioning module is used to determine the real-time position of the detection module; the processing module is used to process the tunnel image and the three-dimensional point cloud data and the real-time position data of the detection module to obtain a tunnel full-line second lining plan of the tunnel lining surface; the identification module is used to identify cracks in the tunnel second lining plan. The positioning module and the detection module can quickly collect data on the tunnel lining surface, and the processing module processes the data to improve the accuracy of the identification module in identifying cracks in the tunnel full-line second lining plan.

Description

Technical Field

[0001] The invention relates to the technical field of tunnel engineering monitoring, and in particular to a tunnel lining surface crack detection device. Background Art

[0002] The existing crack detection work is mainly completed by manual methods. The manual method uses human eye positioning and manual drawing to identify cracks and record their distribution and shape, and uses tools such as vernier calipers and crack microscopes to measure the width of cracks. Practice has shown that manual damage inspection has poor accuracy, strong subjectivity, and is prone to omissions and errors. Summary of the invention

[0003] The purpose of the present invention is to provide a tunnel lining surface crack detection device that can quickly collect tunnel data and accurately identify tunnel cracks. The specific technical solution is as follows:

[0004] The present invention provides a tunnel lining surface crack detection device, comprising a track, a detection module, a positioning module, a processing module and an identification module;

[0005] The track is arranged at the top of the tunnel, and the track is arranged along the direction of the central axis of the tunnel;

[0006] The detection module is slidably arranged on the track, and is used to scan and photograph the tunnel lining surface to obtain a tunnel image and a three-dimensional point cloud respectively;

[0007] The positioning module is arranged at the top central axis of the tunnel entrance, and is used to determine the real-time position of the detection module;

[0008] The processing module is electrically connected to the detection module and the positioning module respectively, and is used to process the tunnel image and the three-dimensional point cloud data and the real-time position data of the detection module to obtain a second lining plane diagram of the entire tunnel lining surface;

[0009] The identification module is electrically connected to the processing module, and the identification module is used to identify cracks in the tunnel secondary lining plan.

[0010] Optionally, the positioning module includes a mounting frame, a rotating element, a laser ranging element and a first electronic compass, the mounting frame is arranged on the top of the tunnel; the rotating element is arranged on the mounting frame, and the output shaft of the rotating element is connected to the laser ranging element; the laser ranging element is used to emit laser to the detection module to measure the distance between the positioning module and the detection module; the first electronic compass is arranged on the mounting frame, and the first electronic compass is used to detect the azimuth and inclination of the positioning module.

[0011] Optionally, the detection module includes a driving element, a mounting body, a detection element, an image positioning element and a second electronic compass, the driving element is slidably disposed on the track, and the mounting body is connected to the driving element;

[0012] The detection element includes a plurality of detection units arranged along the circumference of the mounting body;

[0013] The detection unit includes a mechanical arm, a laser radar and a first camera, wherein the mechanical arm is rotatably connected to the installation body; the laser radar and the first camera are both arranged on the mechanical arm, the laser radar is used to scan the tunnel lining surface, and the first camera is used to photograph the tunnel lining surface within the scanning range of the corresponding laser radar;

[0014] The image positioning element comprises a grid target and a second camera arranged on the mounting body, wherein the grid target is used to receive the laser line emitted by the laser ranging element; the second camera is arranged facing the grid target, and the second camera is used to photograph the grid target;

[0015] The second electronic compass is arranged on the installation body, and is used to detect the azimuth and inclination of the detection module.

[0016] Optionally, the processing module includes a first control component and a second control component, the first control component is arranged on the installation body, and the first control component is electrically connected to the detection module and the identification module respectively; the second control component is arranged on the mounting frame, and the second control component is electrically connected to the positioning module and the first control component respectively.

[0017] The present invention also provides a method for detecting cracks on the surface of a tunnel lining, which uses the tunnel lining surface crack detection device as claimed in claim 4 to detect cracks on the surface of a tunnel lining, and is characterized in that it comprises the following steps:

[0018] S1: Install a tunnel lining surface crack detection device, specifically: divide the entire tunnel into multiple areas, set one detection unit for each area, calibrate the laser radar, the first camera external parameters and the first camera internal parameters, and synchronize the timestamps of the laser radar and the first camera;

[0019] S2: The tunnel lining surface crack detection device is used to scan and photograph the tunnel lining surface in each area to form a color point cloud;

[0020] S3: The color point clouds of each area are registered according to the ICP algorithm to obtain the 3D point cloud of the whole tunnel, and the 3D point cloud of the whole tunnel is expanded according to the point cloud expansion algorithm to obtain the 2D point cloud, and then the 2D point cloud is projected to obtain the second lining plan of the whole tunnel;

[0021] S4: After cutting and filling the second lining plan of the whole tunnel, the plan is input into the crack detection model based on the U-Net algorithm to identify the cracks in the second lining plan of the whole tunnel, and obtain the pixel coordinates, length and width of the cracks;

[0022] S5: The identified cracks are reproduced in the second lining plan of the entire tunnel according to the relationship between pixel coordinates, and the positions of the cracks in the two-dimensional point cloud are inversely calculated according to the two-dimensional point cloud projection algorithm, and then the positions of the cracks in the three-dimensional point cloud are inversely calculated according to the point cloud expansion algorithm.

[0023] Optionally, in S1, a scanning field of view angle of a laser radar is greater than 70°, and a repeated scanning range of two adjacently arranged laser radars is 10%-30%; and a shooting range of the first camera covers a scanning range of the corresponding laser radar;

[0024] The S2 includes:

[0025] S2.1, start the driving element of the detection module to move the detection module along the track, use the laser radar to scan the tunnel lining surface of the area to form a three-dimensional point cloud, and at the same time, the first camera shoots the tunnel lining surface of each area to obtain the tunnel image of each area;

[0026] S2.2. Construct a local coordinate system, and adjust the point cloud coordinates obtained by the laser radar scanning to the construction coordinates under the local coordinate system in real time based on the local coordinate system. The local coordinate system takes the center of the positioning module as the origin O, the vertical direction as the Z axis, the tunnel centerline direction as the Y axis, and the straight line direction perpendicular to the YOZ plane and passing through the origin as the X axis. Specifically:

[0027] S2.2.1. Calculate the emission point coordinates of the laser distance measuring element based on the fixed installation distance (a 0 ,b 0 ,c 0 ), and calculate the local coordinates of the laser landing point according to the coordinates of the emission point (a 1 ,b 1 ,c 1 ), the specific calculation formula is as follows:

[0028]

[0029] Among them: 1 is the distance between the grid target and the positioning module measured by the laser ranging element; α 1 is the real-time azimuth measured by the first electronic compass, β 1 is the inclination angle measured by the first electronic compass, α 0 The initial azimuth in the Y-axis direction is measured by the first electronic compass;

[0030] S2.2.2. Using a second camera to obtain a grid target image, wherein the grid target image is composed of white, green and red;

[0031] S2.2.3, calculate the actual difference between the laser point and the center point of the grid target;

[0032] S2.2.4. Calculate the coordinates of the center point on the grid target (a) based on the azimuth and inclination measured by the second electronic compass. 2 ,b 2 ,c 2 ), the specific calculation formula is as follows:

[0033]

[0034] Where: Δj is the vertical pixel coordinate difference between the laser point and the center point of the grid target; Δk is the horizontal pixel coordinate difference between the laser point and the center point of the grid target; is the average value of the actual size of a single pixel; α 2 is the azimuth measured by the second electronic compass, β 2 The inclination angle measured by the second electronic compass;

[0035] S2.2.5. Calculate the coordinates of the laser radar and the second camera respectively according to the fixed distance between the laser radar and the second camera, and convert the point cloud coordinates scanned by each laser radar into the construction coordinates in the local coordinate system;

[0036] S2.3, projecting the construction coordinates of the three-dimensional point cloud into the UV coordinate system of the first camera according to the external parameters of the first camera, and obtaining the pixel value of each point in the three-dimensional point cloud at the corresponding position of the corresponding tunnel image;

[0037] S2.4. Assign pixel values ​​to each point to form a colored three-dimensional point cloud for each area.

[0038] Optionally, the S3 includes:

[0039] S3.1, using statistical filtering algorithm to remove discrete points in the three-dimensional point cloud;

[0040] S3.2. Use the ICP algorithm to register the color point clouds of each area to obtain the three-dimensional point cloud of the entire tunnel, and use the octree algorithm to traverse and retrieve the three-dimensional point cloud of the entire tunnel to remove duplicate points;

[0041] S3.3. Expand the 3D point cloud of the entire tunnel according to the tunnel section contour line according to the point cloud expansion algorithm to obtain a 2D point cloud, specifically:

[0042] The coordinates (x′, y′, z′) of the straight line segment point cloud in the three-dimensional point cloud are converted to the coordinates (x, y, z) of the XOY plane. The specific calculation formula is as follows:

[0043]

[0044] Where: G is the rotation matrix, T is the translation matrix;

[0045] The point cloud coordinates (x i ,y i ,z i ) According to the cylindrical formula formed by the arc, the point cloud coordinates (x i ′,y i ′,z i ′), the specific calculation formula is as follows:

[0046]

[0047] Where: θ i For any point (x i ,y i ,z i ) with the axis at the center of the circle and the angle between the node and the node plane formed by the axis at the center of the circle. The node is the intersection of the arc segment with other arc segments and straight line segments (x 1 ,y 1 ,z 1 ); is the angle between the tunnel centerline and the XOZ plane; (x 0 ,y 0 ,z 0 ) is the coordinate of the axis at the center of the circle in a certain tunnel cross section.

[0048] S3.4. Project the 2D point cloud onto the XOY plane to obtain the second lining plan of the entire tunnel, specifically:

[0049] S3.4.1. Design the tunnel plan in the XOY plane based on the tunnel centerline and the perimeter of the second lining of the section;

[0050] S3.4.2. Divide the tunnel plane into a grid of 1 mm × 1 mm;

[0051] S3.4.3, set the ordinate in the two-dimensional point cloud to 0, and then project the point cloud onto the tunnel plane empty map;

[0052] S3.4.4. Count the pixel values ​​of the point cloud in each grid and take the average value as the pixel value of the grid;

[0053] S3.4.5. Each grid cell forms a pixel point, forming a second lining plan of the entire tunnel.

[0054] Optionally, the S4 includes:

[0055] S4.1. First, use labelme software to process the lining crack images of other tunnels and produce crack label images of set size; then use the U-Net algorithm to train and learn the produced crack label images to establish a tunnel crack semantic segmentation model.

[0056] S4.2, cropping the full tunnel second lining plan in S3.4.5 according to the set size input by U-Net to obtain a cropped compliance image, filling the insufficient size part with blank pixels, and then numbering the cropped compliance images one by one according to the distance from the starting point and from the left side of the tunnel to the right side of the tunnel;

[0057] S4.3, input the cropped compliance image into the tunnel crack semantic segmentation model for recognition, and extract its mask to obtain the corresponding pixel coordinates;

[0058] S4.4. All the recognized images are stitched together according to the numbers to form a tunnel-wide crack recognition result map, and the pixel coordinates of each crack in the tunnel-wide crack recognition result map are calculated.

[0059] The present invention also provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the tunnel lining surface crack detection method as described above is implemented.

[0060] The present invention also provides an electronic device, comprising: at least one processor, at least one memory and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, the tunnel lining surface crack detection method as described above is performed.

[0061] By applying the technical solution of the present invention, the detection module moves on the track at the top of the tunnel while scanning the tunnel lining surface to obtain a three-dimensional point cloud of the tunnel lining surface, and the positioning module can determine the real-time position of the detection module, and the processing module processes the three-dimensional point cloud data of the tunnel lining surface and the real-time position data of the detection module to obtain the second lining plan of the entire tunnel, and finally the recognition module recognizes cracks in the second lining plan of the entire tunnel. In the present invention, the positioning module and the detection module can quickly collect data on the tunnel lining surface, and the processing module processes the data to improve the accuracy of the recognition module in recognizing cracks in the second lining plan of the entire tunnel.

[0062] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0064] Figure 1 is a schematic diagram of the structure of a tunnel lining surface crack detection device according to an embodiment of the present invention;

[0065] Figure 2 is a schematic structural diagram of a detection element in an embodiment of the present invention;

[0066] Figure 3 is a schematic structural diagram of an image positioning element in an embodiment of the present invention;

[0067] Figure 4 is a schematic diagram of a grid target and a laser line in an embodiment of the present invention;

[0068] Figure 5 is a schematic diagram of the structure after the track, driving element and tunnel lining surface are installed in an embodiment of the present invention;

[0069] Figure 6 is a schematic diagram of a three-dimensional point cloud of the entire tunnel in an embodiment of the present invention;

[0070] Figure 7 It is a plan view of the second lining of the entire tunnel in the embodiment of the present invention;

[0071] Figure 8 This is a diagram showing the crack identification results for the entire tunnel in an embodiment of the present invention.

[0072] Fig. 9 It is a cut-out schematic diagram of the second lining plan of the entire tunnel in S4.2 in the embodiment of the present invention.

[0073] Description of Figure Numbers:

[0074] 1 Track, 1.1 Pillar, 1.2 Guide body, 2 Detection module, 2.1 Driving element, 2.1.1 Roller, 2.1.2 Fixed bearing, 2.1.3 Gear, 2.1.4 Transmission chain, 2.1.5 Driving member, 2.1.6 Transmission member, 2.2 Mounting body, 2.3 Detection element, 2.3.1 Detection unit, 2.3.1.1 Mechanical arm, 2.3.1.2 Laser radar, 2.3.1.3 First camera, 2.4 Image positioning element, 2.4.1 Grid target, 2.4.2 Second camera, 2.4.3 Shooting range, 2.5 Second electronic compass, 2.6 Light strip, 3 Positioning module, 3.1 Mounting frame, 3.2 Rotating element, 3.3 Laser ranging element, 3.3.1 Laser line, 3.5 First electronic compass, 4 Processing module, 4.1 First control component, 4.2 Second control component, 5 Tunnel lining surface. DETAILED DESCRIPTION

[0075] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0076] In one embodiment, if Figure 1 As shown, a tunnel lining surface crack detection device includes a track 1, a detection module 2, a positioning module 3, a processing module 4 and an identification module; the track 1 is arranged on the top of the tunnel, and the track 1 is arranged along the direction of the tunnel center axis; the detection module 2 is slidably arranged on the track 1, and is used to scan and shoot the tunnel lining surface 5 to obtain a tunnel image and a three-dimensional point cloud respectively; the positioning module 3 is arranged at the top center axis of the tunnel entrance, and the positioning module 3 is used to determine the real-time position of the detection module 2; the processing module 4 is electrically connected to the detection module 2 and the positioning module 3 respectively, and the processing module 4 is used to process the tunnel image and the three-dimensional point cloud data and the real-time position data of the detection module 2 to obtain the tunnel full-line second lining plan view of the tunnel lining surface 5; the identification module is electrically connected to the processing module 4, and the identification module is used to identify cracks in the tunnel second lining plan view. In the actual detection operation, while the detection module 2 moves on the track 1 at the top of the tunnel, it scans the tunnel lining surface 5 to obtain the three-dimensional point cloud of the tunnel lining surface 5, and the positioning module 3 can determine the real-time position of the detection module 2, and the three-dimensional point cloud data of the tunnel lining surface 5 and the real-time position data of the detection module 2 are processed by the processing module 4 to obtain the second lining plan of the entire tunnel, and the recognition module is provided with a tunnel crack semantic segmentation model, which can identify cracks in the second lining plan of the entire tunnel. In this embodiment, the positioning module 3 and the detection module 2 can quickly collect data on the tunnel lining surface 5, and the processing module 4 processes the data to improve the accuracy of the recognition module in identifying cracks in the second lining plan of the entire tunnel.

[0077] Based on the above embodiment, the positioning module 3 includes a mounting frame 3.1, a rotating element 3.2, a laser distance measuring element 3.3 and a first electronic compass 3.5. The mounting frame 3.1 is arranged on the top of the tunnel; the rotating element 3.2 is arranged on the mounting frame 3.1, and the output shaft of the rotating element 3.2 is connected to the laser distance measuring element 3.3; the laser distance measuring element 3.3 is used to emit laser to the detection module 2 to measure the distance between the positioning module 3 and the detection module 2; the first electronic compass 3.5 is arranged on the mounting frame 3.1, and the first electronic compass 3.5 is used to detect the azimuth and inclination of the positioning module 3. The rotating element 3.2 can drive the laser distance measuring element 3.3 to rotate, so that the laser emitted by the laser distance measuring element 3.3 can be aligned with the detection module 2, thereby measuring the distance between the positioning module 3 and the detection module 2, and the first electronic compass 3.5 is used to detect the azimuth and inclination of the positioning module 3, which are used for the subsequent calculation of the coordinates of the laser landing point.

[0078] In one embodiment, in conjunction with Figures 2 to 4 The detection module 2 includes a driving element 2.1, a mounting body 2.2, a detection element 2.3, an image positioning element 2.4 and a second electronic compass 2.5. The driving element 2.1 is slidably arranged on the track 1, and the mounting body 2.2 is connected to the driving element 2.1; the detection element 2.3 includes a plurality of detection units 2.3.1 arranged along the circumference of the mounting body 2.2; the detection unit 2.3.1 includes a mechanical arm 2.3.1.1, a laser radar 2.3.1.2 and a first camera 2.3.1.3, and the mechanical arm 2.3.1.1 is rotatably connected to the mounting body 2.2; the laser radar 2.3.1.2 and the first camera 2.3.1.3 are both arranged on the mechanical arm 2.3.1.1, and the laser radar The laser radar 2.3.1.2 is used to scan the tunnel lining surface 5, and the first camera 2.3.1.3 is used to photograph the tunnel lining surface 5 within the scanning range of the corresponding laser radar 2.3.1.2; the image positioning element 2.4 includes a grid target 2.4.1 and a second camera 2.4.2 arranged on the installation body 2.2, the grid target 2.4.1 is used to receive the laser line 3.3.1 emitted by the laser ranging element 3.3; the second camera 2.4.2 is arranged opposite to the grid target 2.4.1, and the second camera 2.4.2 is used to photograph the grid target 2.4.1; the second electronic compass 2.5 is arranged on the installation body 2.2, and the second electronic compass 2.5 is used to detect the azimuth and inclination of the detection module 2. By rotating the mechanical arm 2.3.1.1 of each detection unit 2.3.1, the distance between two adjacent detection units 2.3.1 can be adjusted, so as to ensure that the scanning and shooting range 2.4.3 of the detection element 2.3 can cover the lining surface within the tunnel section, and the driving element 2.1 can drive the detection element 2.3 to move along the track 1 through the installation body 2.2, so as to scan and shoot the lining surface of the entire tunnel, the grid target 2.4.1 can receive the laser line 3.3.1 emitted by the laser ranging element 3.3, and the grid target 2.4.1 image shot by the second camera 2.4.2, and then the grid target 2.4.1 image is processed by the processing module 4 to obtain the center coordinates of the grid target 2.4.1, and the data measured by the detection element 2.3 and the positioning module 3 are processed to obtain the second lining plan of the entire tunnel.

[0079] like Figure 5As shown, the track 1 includes two relatively arranged monorails, the monorails include connected pillars 1.1 and guide bodies 1.2, the pillars 1.1 are connected to the tunnel lining surface 5, and the guide bodies 1.2 are provided with slide grooves, and the slide grooves of the two guide bodies 1.2 are relatively arranged; the driving element 2.1 includes a base, a driving member 2.1.5, a transmission member 2.1.6 and two rolling assemblies, the driving member 2.1.5 is arranged on the base, and the rolling assembly includes a fixed bearing 2.1.2, a roller 2.1.1, a gear 2.1.3 and a transmission chain 2.1.4, the fixed bearing 2.1.2 is arranged at the bottom of the base, and rollers 2.1.1 are arranged at both ends of the fixed bearing 2.1.2. .1, two rollers 2.1.1 are respectively set in the slide grooves of two monorails for rolling, the gear 2.1.3 is set on the fixed bearing 2.1.2, and the gears 2.1.3 in the two rolling components are driven by the transmission chain 2.1.4, the output shaft of the driving member 2.1.5 is connected with a rolling component through the transmission member 2.1.6, the driving member 2.1.5 drives the rolling component to move along the track 1, and the rolling component drives another rolling component to move through the transmission chain 2.1.4, and the driving member 2.1.5 is electrically connected to the first control component 4.1, and the first control component 4.1 can control the moving speed of the driving element 2.1 through the driving member 2.1.5.

[0080] Preferably, the detection element 2.3 further comprises a light strip 2.6, both ends of which are respectively connected to two adjacent mechanical arms 2.3.1.1, and the light strip 2.6 passes through each mechanical arm 2.3.1.1 in sequence and is arranged in a ring shape for illumination, so as to facilitate scanning and photographing tunnel images.

[0081] In addition, the processing module 4 includes a first control component 4.1 and a second control component 4.2. The first control component 4.1 is arranged on the installation body 2.2, and the first control component 4.1 is electrically connected to the detection module 2 and the identification module respectively; the second control component 4.2 is arranged on the mounting frame 3.1, and the second control component 4.2 is electrically connected to the positioning module 3 and the first control component 4.1 respectively. The second control component 4.2 can calculate the coordinates of the laser landing point through the direction angle, inclination angle and laser ranging structure data measured by the positioning module 3, and transmit the calculation result to the first control component 4.1, and the first control component 4.1 can process the calculation result and the relevant data measured by the detection module 2 to obtain the second lining plan of the whole tunnel, and input the second lining plan of the whole tunnel into the identification module for identification. Through the data processing of the first control component 4.1 and the second control component 4.2, the accuracy of the identification module in identifying cracks in the second lining plan of the whole tunnel can be effectively improved.

[0082] The present invention also provides a method for detecting cracks on the surface of a tunnel lining, which uses the above-mentioned device for detecting cracks on the surface of a tunnel lining to detect cracks on the surface of a tunnel lining, and comprises the following steps:

[0083] S1: Install a tunnel lining surface crack detection device, specifically: divide the entire tunnel into multiple areas, set up a detection unit 2.3.1 for each area, calibrate the external parameters of the laser radar 2.3.1.2, the first camera 2.3.1.3 and the internal parameters of the first camera 2.3.1.3, and synchronize the timestamps of the laser radar 2.3.1.2 and the first camera 2.3.1.3.

[0084] Among them: the scanning angle of a laser radar 2.3.1.2 is greater than 70°, and the repeated scanning range of two adjacent laser radars 2.3.1.2 is 10%-30%; the shooting range 2.4.3 of the first camera 2.3.1.3 covers the scanning range of the corresponding laser radar 2.3.1.2; it is also necessary to ensure that the positioning module 3 can be positioned to the detection module 2, that is, to ensure that the laser pointer of the positioning module 3 falls on the center point of the grid target 2.4.1 of the image positioning element 2.4 of the detection module 2; then adjust the laser radar 2.3.1.2 and the first camera 2.3.1.3 positions of the detection unit 2.3.1 corresponding to each area, and calibrate the camera external parameters and camera internal parameters of the laser radar 2.3.1.2 and the first camera 2.3.1.3 of the detection unit 2.3.1 in each area, specifically:

[0085] Arrange multiple chessboard calibration plates on the vault, waist, and spandrel of the secondary lining at the entrance. The chessboard calibration plates have the same size and the same number of chessboard grids to ensure that one laser radar 2.3.1.2 or the first camera 2.3.1.3 can scan at least 3 calibration plates. Then, the Zhang Youding calibration method is used to calculate the internal parameter matrix and distortion coefficient of the first camera 2.3.1.3, that is, mark the corner points in the tunnel image taken by the first camera 2.3.1.3, measure the actual physical coordinates of the corresponding corner points according to the total station, and solve the internal parameter matrix and distortion coefficient according to the corresponding relationship using the least squares method; then, mark the corner points corresponding to the first camera 2.3.1.3 in the point cloud data scanned by the laser radar 2.3.1.2, and then solve the rotation matrix R between the camera coordinate system and the radar coordinate system. o and the translation vector t o ,Right now:

[0086] P cam =R o *P lidar +t o ;

[0087] P cam is the coordinate of the corner point in the camera coordinate system; P lidar are the coordinates of the corner point in the LiDAR coordinate system.

[0088] This enables joint calibration of the laser radar 2.3.1.2 and the first camera 2.3.1.3.

[0089] Then adjust the time of the first camera 2.3.1.3 and the laser radar 2.3.1.2, that is, start the first camera 2.3.1.3 shooting and the laser radar 2.3.1.2 scanning simultaneously to synchronize their timestamps.

[0090] S2: using a tunnel lining surface crack detection device to scan and photograph the tunnel lining surface 5 of each area to form a color point cloud;

[0091] Start the detection module 2 and run it at a constant and low speed on the track, synchronize the measurement time of the positioning module 3, start scanning and photographing the tunnel lining surface 5, when the laser radar 2.3.1.2 scans the tunnel to form a three-dimensional point cloud, the positioning module 3 determines the accurate position of the laser radar 2.3.1.2 in real time, adjusts the point cloud space coordinates of the laser radar 2.3.1.2 in real time, projects the point cloud data into the UV coordinate system of the first camera 2.3.1.3 according to the camera external parameters, obtains the pixel value of each point at the corresponding position of the tunnel, and assigns it to the corresponding point in the point cloud to form a high-precision color three-dimensional point cloud, specifically:

[0092] S2 includes:

[0093] S2.1, start the driving element 2.1 of the detection module 2, so that the detection module 2 moves along the track 1, and use the laser radar 2.3.1.2 to scan the tunnel lining surface 5 in the area to form the following Figure 6 The three-dimensional point cloud shown, at the same time, the first camera 2.3.1.3 shoots the tunnel lining surface 5 of each area to obtain the tunnel image of each area; in this embodiment, the running speed of the detection module is less than or equal to 10km / h.

[0094] S2.2. Construct a local coordinate system. Use the local coordinate system as the standard to adjust the point cloud coordinates obtained by scanning with the laser radar 2.3.1.2 to the construction coordinates under the local coordinate system in real time. The local coordinate system takes the center of the positioning module 3 as the origin O, the vertical direction as the Z axis, the tunnel centerline direction as the Y axis, and the straight line direction perpendicular to the YOZ plane and passing through the origin as the X axis. Specifically:

[0095] S2.2.1. Calculate the emission point coordinates of the laser distance measuring element 3.3 according to the fixed installation distance (a 0 ,b 0 ,c 0 ), and calculate the local coordinates of the laser landing point according to the coordinates of the emission point (a 1 ,b 1 ,c 1 ), the specific calculation formula is as follows:

[0096]

[0097] Among them: 1 is the distance between the grid target 2.4.1 and the positioning module 3 measured by the laser distance measuring element 3.3; α 1 is the real-time azimuth measured by the first electronic compass 3.5, β 1 is the inclination angle measured by the first electronic compass 3.5, α 0 The initial azimuth in the Y-axis direction is measured by the first electronic compass 3.5;

[0098] S2.2.2. Use the second camera 2.4.2 to obtain an image of the grid target 2.4.1, wherein the image of the grid target 2.4.1 is composed of white, green and red;

[0099] S2.2.3, calculate the actual difference between the laser point and the center point of the grid target 2.4.1;

[0100] First, the HSV spatial threshold algorithm is used to extract green pixels and red pixels respectively, and they are converted into the grid target 2.4.1 contour binarization image and the laser point binarization image respectively. Then, the Zhang-Suen algorithm is used to refine the grid target 2.4.1 binarization image until the pixel width of the grid line is a single pixel. The vertical center line and the horizontal center line are calculated using the cumulative Hough transform algorithm. Since the set grid size is fixed, the corresponding single pixel actual size can be calculated as the grid line actual length / grid line pixel length, and the average value of the single pixel actual size is obtained. Then calculate the intersection of the center lines of the grid, and use this point as the center point of the grid target 2.4.1 and the origin of the local pixel coordinate system; use the Hough gradient method to fit the inner and outer circles, from which the center point and pixel radius of the circle can be calculated. Given that the radius of the inner and outer circles is fixed, use the Hough gradient method to fit the pixel circle of the laser point, and calculate the pixel coordinates of the center point of the circle. From this, it can be calculated that the actual difference between the laser point and the center point of the grid target is Δj is the vertical pixel coordinate difference between the two points, and Δj is the horizontal pixel coordinate difference between the two points.

[0101] S2.2.4. Calculate the coordinates of the center point of grid target 2.4.1 (a 2 ,b 2 ,c 2 ), the specific calculation formula is as follows:

[0102]

[0103] Where: Δj is the vertical pixel coordinate difference between the laser point and the center point of the grid target; Δk is the horizontal pixel coordinate difference between the laser point and the center point of the grid target; is the average value of the actual size of a single pixel; α2 is the azimuth angle measured by the second electronic compass 2.5, β 2 The inclination angle measured by the second electronic compass 2.5;

[0104] S2.2.5. Calculate the coordinates of the laser radar 2.3.1.2 and the second camera 2.4.2 respectively according to the fixed distance between the laser radar 2.3.1.2 and the second camera 2.4.2, and convert the point cloud coordinates scanned by each laser radar 2.3.1.2 into the construction coordinates in the local coordinate system;

[0105] After all tunnel inspections are completed, the rotation matrix R and translation vector t between the camera coordinate system and the radar coordinate system are used to project the point cloud into the camera's UV coordinate system, obtain the pixel value of each point at the corresponding position of the camera image and assign it to the corresponding point, thus forming point cloud data with XYZ coordinates and RGB color values. Specifically:

[0106] S2.3, projecting the construction coordinates of the three-dimensional point cloud into the UV coordinate system of the first camera 2.3.1.3 according to the external parameters of the first camera 2.3.1.3, and obtaining the pixel value of each point in the three-dimensional point cloud at the corresponding position of the corresponding tunnel image;

[0107] S2.4. Assign pixel values ​​to each point to form a colored three-dimensional point cloud for each area.

[0108] S3: The color point clouds of each area are registered according to the ICP algorithm to obtain the 3D point cloud of the entire tunnel, and the 3D point cloud of the entire tunnel is expanded according to the point cloud expansion algorithm to obtain a 2D point cloud, and then the 2D point cloud is projected to obtain the following: Figure 7 The plan view of the second lining along the entire length of the tunnel is shown;

[0109] S3 includes:

[0110] S3.1, using statistical filtering algorithm to remove discrete points in the three-dimensional point cloud;

[0111] S3.2. Use the ICP algorithm to register the color point clouds of each area to the three-dimensional point cloud of the entire tunnel, and use the octree algorithm to traverse and retrieve the three-dimensional point cloud of the entire tunnel to remove duplicate points;

[0112] S3.3. Expand the 3D point cloud of the entire tunnel according to the tunnel section contour line according to the point cloud expansion algorithm to obtain a 2D point cloud, specifically:

[0113] The three-dimensional point cloud data of the straight line segment is expanded into two-dimensional point cloud data, which can be expanded according to the Euclidean transformation: the normal vector of the plane where the straight line segment point cloud is located is known to be: Ax+By+Cz+D=0, A, B, C, D are unknown coefficients; the normal vector of the XOY plane is (0,0,1), so to offset the point cloud data to the XOY plane, its normal vector needs to be rotated to (0,0,1), where the rotation matrix G is:

[0114]

[0115] The translation matrix T can be calculated based on the intersection points (x 1 ,y 1 ,z 1 ), and the coordinates of the corresponding intersection points after the offset of other point cloud data are known to be (x 1 ′,y 1 ′,0), from which the translation T can be calculated 1 ′-x 1 ,y 1 ′-y 1 ,-z 1 );

[0116] The coordinates (x′, y′, z′) of the straight line segment point cloud in the three-dimensional point cloud are converted to the coordinates (x, y, z) of the XOY plane. The specific calculation formula is as follows: T = (x 1 ′-x 1 ,y 1 ′-y 1 ,-z 1 )

[0117]

[0118] Where: G is the rotation matrix, T is the translation matrix;

[0119] For the point cloud data of the arc segment, first expand all the point cloud data to form a two-dimensional point cloud, and then use the Euclidean transformation algorithm to transform it to the XOY plane. The following provides a method for expanding to form a two-dimensional point cloud:

[0120] The point cloud coordinates (x i ,y i ,z i ) According to the cylindrical formula formed by the arc, the point cloud coordinates (x i ′,y i ′,z i ′), the specific calculation formula is as follows:

[0121]

[0122] Where: θ i For any point (x i ,y i ,z i ) with the axis at the center of the circle and the angle between the node and the node plane formed by the axis at the center of the circle. The node is the intersection of the arc segment with other arc segments and straight line segments (x 1 ,y 1 ,z 1 ); is the angle between the tunnel centerline and the XOZ plane; (x 0 ,y 0 ,z 0 ) is the coordinate of the axis at the center of the circle in a certain tunnel cross section.

[0123] S3.4. Project the 2D point cloud onto the XOY plane to obtain the second lining plan of the entire tunnel, specifically:

[0124] The formed two-dimensional point cloud model is projected to the XOY plane, that is, firstly, the tunnel plane map is designed in the XOY plane according to the tunnel central axis and the perimeter of the second lining of the section, and then the plane map is divided into 1mm*1mm grids. After that, the vertical coordinate in the two-dimensional point cloud model is set to 0, and the point cloud can be projected to the tunnel plane map. The pixel values ​​of the point cloud in each grid are counted, and the average value is taken as the pixel value of the grid. Each grid constitutes a pixel point to form the second lining plane map of the entire tunnel.

[0125] S3.4.1. Design the tunnel plan in the xoy plane based on the tunnel centerline and the perimeter of the second lining of the section;

[0126] S3.4.2. Divide the tunnel plane into a grid of 1 mm × 1 mm;

[0127] S3.4.3, set the ordinate in the two-dimensional point cloud to 0, and then project the point cloud onto the tunnel plane empty map;

[0128] S3.4.4. Count the pixel values ​​of the point cloud in each grid and take the average value as the pixel value of the grid;

[0129] S3.4.5. Each grid cell forms a pixel point, forming a second lining plan of the entire tunnel.

[0130] S4: After cutting and filling the second lining plan of the whole tunnel, the plan is input into the crack detection model based on the U-Net algorithm to identify the cracks in the second lining plan of the whole tunnel, and obtain the pixel coordinates, length and width of the cracks;

[0131] S4 includes:

[0132] S4.1. First, use labelme software to process the tunnel lining crack images of other tunnels to produce crack label images of set size; then use U-Net algorithm to train and learn the produced crack label images to establish a tunnel crack semantic segmentation model. In this embodiment, the tunnel lining crack images of other tunnels can be obtained from a public lining crack dataset, and the number is more than 1,000, and the set size is 512×512.

[0133] S4.2, crop the full tunnel second lining plan in S3.4.5 according to the set size input by U-Net to obtain the cropped compliance image, fill the insufficient size part with blank pixels, and then number the cropped compliance images one by one according to the distance from the starting point and from the left side of the tunnel to the right side of the tunnel; Fig. 9 As shown, the second lining plan view of the entire tunnel in S3.4.5 is cut and divided into r×s images. Fig. 9 The r in the equation is the number of images cropped along the central axis of the tunnel. Fig. 9 The s in the equation is the number of images cropped along the tunnel cross section expansion direction. Fig. 9 The direction indicated by the arrow in is the extension direction of the tunnel, and the coordinates of the starting pixel point in the upper left corner of the A1 image are set to (0,0).

[0134] S4.3. Input the cropped compliant image into the tunnel crack semantic segmentation model for recognition, and extract its mask to obtain the corresponding pixel coordinates (u1, v1); the tunnel crack semantic segmentation model is set in the recognition module.

[0135] S4.4, all the recognized images are stitched together according to the numbers to form Figure 8 The result map of crack identification along the entire tunnel is shown, and the pixel coordinates of each crack in the result map of crack identification along the entire tunnel are calculated. The (x, y) coordinates of the crack can be calculated according to the pixel block size. The calculation method is as follows:

[0136] In this embodiment, assuming that the identified image is the mth row and nth column, that is, the coordinates of the upper left corner pixel of the image are (512×m+1, 512×n+1), then the coordinates of the pixel corresponding to the crack are (512m+1+u1, 512×n+1+v1).

[0137] S5: Reproduce the identified cracks into the second lining plan of the whole tunnel according to the relationship between pixel coordinates.

[0138] Since the grid size is 1mm×1mm, the size of each pixel is 1mm×1mm, so the horizontal and vertical coordinates (x', y') of the crack pixel can be calculated as: ((512×m+1+u1)mm, (512×n+1+v1)mm)

[0139] The corresponding two-dimensional point cloud coordinates are found according to the horizontal and vertical coordinates of the pixel point, thereby obtaining the z' coordinate of the point cloud.

[0140] The position of the crack in the two-dimensional point cloud is calculated based on the two-dimensional point cloud projection algorithm, and then the position of the crack in the three-dimensional point cloud is calculated based on the point cloud expansion algorithm. The specific calculation formula is as follows:

[0141]

[0142] The tunnel lining surface crack detection method in this embodiment can convert the three-dimensional point cloud model of the entire tunnel into an image, identify the cracks in the lining crack image of the entire tunnel, and reproduce them in the three-dimensional point cloud model. The detection method in this embodiment can identify hidden cracks in the tunnel, and the tunnel lining surface crack detection device can speed up the data collection process, effectively improve the point cloud collection density, form a high-definition point cloud image, realize automatic and accurate identification and positioning of tunnel cracks, effectively improve the efficiency of tunnel crack operation and maintenance, and accurately present the health of the tunnel secondary lining. In addition, the accuracy of crack recognition through images is higher, and the problem of long recognition time caused by point cloud big data is effectively avoided.

[0143] This embodiment also provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method for detecting cracks on the surface of a tunnel lining as described above is implemented.

[0144] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0145] This embodiment also provides an electronic device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory. When the computer program instructions are executed by the processor, a method for detecting cracks on the surface of a tunnel lining as described in any one of the above items is performed.

[0146] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0147] The electronic device may be a computing device such as a mobile phone, a desktop computer, a notebook, a PDA, a cloud server, etc. The electronic device may include, but is not limited to, a processor and a memory. For example, the electronic device may also include an input / output device, a network access device, a bus, etc.

[0148] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0149] The memory can be used to store the computer program and / or module, and the processor implements the computer program by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0150] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A tunnel lining surface crack detection device, characterized in that: The tunnel lining surface crack detection device comprises a track (1), a detection module (2), a positioning module (3), a processing module (4) and an identification module; The track (1) is arranged at the top of the tunnel, and the track (1) is arranged along the direction of the central axis of the tunnel; The detection module (2) is slidably arranged on the track (1) and is used to scan and photograph the tunnel lining surface (5) to obtain a tunnel image and a three-dimensional point cloud respectively; The positioning module (3) is arranged at the top central axis of the tunnel entrance, and the positioning module (3) is used to determine the real-time position of the detection module (2); The processing module (4) is electrically connected to the detection module (2) and the positioning module (3) respectively, and the processing module (4) is used to process the tunnel image and three-dimensional point cloud data as well as the real-time position data of the detection module (2) to obtain a second lining plane diagram of the entire tunnel lining surface (5); The identification module is electrically connected to the processing module (4), and the identification module is used to identify cracks in the tunnel secondary lining plan; The positioning module (3) comprises a mounting frame (3.1), a rotating element (3.2), a laser distance measuring element (3.3) and a first electronic compass (3.5); the mounting frame (3.1) is arranged on the top of the tunnel; the rotating element (3.2) is arranged on the mounting frame (3.1), and the output shaft of the rotating element (3.2) is connected to the laser distance measuring element (3.3); the laser distance measuring element (3.3) is used to emit laser light to the detection module (2) to measure the distance between the positioning module (3) and the detection module (2); the first electronic compass (3.5) is arranged on the mounting frame (3.1), and the first electronic compass (3.5) is used to detect the azimuth and inclination of the positioning module (3); The detection module (2) comprises a driving element (2.1), a mounting body (2.2), a detection element (2.3), an image positioning element (2.4) and a second electronic compass (2.5); the driving element (2.1) is slidably arranged on the track (1); and the mounting body (2.2) is connected to the driving element (2.1); The detection element (2.3) comprises a plurality of detection units (2.3.1) arranged along the circumference of the mounting body (2.2); The detection unit (2.3.1) comprises a mechanical arm (2.3.1.1), a laser radar (2.3.1.2) and a first camera (2.3.1.3); the mechanical arm (2.3.1.1) is rotatably connected to the installation body (2.2); the laser radar (2.3.1.2) and the first camera (2.3.1.3) are both arranged on the mechanical arm (2.3.1.1); the laser radar (2.3.1.2) is used to scan the tunnel lining surface (5); and the first camera (2.3.1.3) is used to photograph the tunnel lining surface (5) within the scanning range of the corresponding laser radar (2.3.1.2); The image positioning element (2.4) comprises a grid target (2.4.1) and a second camera (2.4.2) arranged on the mounting body (2.2); the grid target (2.4.1) is used to receive the laser line (3.3.1) emitted by the laser distance measuring element (3.3); the second camera (2.4.2) is arranged facing the grid target (2.4.1), and the second camera (2.4.2) is used to photograph the grid target (2.4.1); The second electronic compass (2.5) is arranged on the installation body (2.2), and the second electronic compass (2.5) is used to detect the azimuth and inclination of the detection module (2).

2. The tunnel lining surface crack detection device according to claim 1, characterized in that: The processing module (4) comprises a first control component (4.1) and a second control component (4.2); the first control component (4.1) is arranged on the mounting body (2.2), and the first control component (4.1) is electrically connected to the detection module (2) and the identification module respectively; the second control component (4.2) is arranged on the mounting frame (3.1), and the second control component (4.2) is electrically connected to the positioning module (3) and the first control component (4.1) respectively.

3. A method for detecting cracks on the surface of a tunnel lining, using the device for detecting cracks on the surface of a tunnel lining as claimed in claim 2 to detect cracks on the surface of a tunnel lining (5), characterized in that: The steps include: S1: Install a tunnel lining surface crack detection device, specifically: divide the entire tunnel into multiple areas, set a detection unit (2.3.1) for each area, calibrate the external parameters of the laser radar (2.3.1.2), the first camera (2.3.1.3) and the internal parameters of the first camera (2.3.1.3), and synchronize the timestamps of the laser radar (2.3.1.2) and the first camera (2.3.1.3); S2: using a tunnel lining surface crack detection device to scan and photograph the tunnel lining surface of each area (5) to form a color point cloud; S3: The color point clouds of each area are registered according to the ICP algorithm to obtain the 3D point cloud of the whole tunnel, and the 3D point cloud of the whole tunnel is expanded according to the point cloud expansion algorithm to obtain the 2D point cloud, and then the 2D point cloud is projected to obtain the second lining plan of the whole tunnel; S4: After cutting and filling the second lining plan of the whole tunnel, the plan is input into the crack detection model based on the U-Net algorithm to identify the cracks in the second lining plan of the whole tunnel, and obtain the pixel coordinates, length and width of the cracks; S5: The identified cracks are reproduced in the second lining plan of the entire tunnel according to the relationship between pixel coordinates, and the positions of the cracks in the two-dimensional point cloud are inversely calculated according to the two-dimensional point cloud projection algorithm, and then the positions of the cracks in the three-dimensional point cloud are inversely calculated according to the point cloud expansion algorithm.

4. The method for detecting cracks on the tunnel lining surface according to claim 3, characterized in that: In S1, the scanning angle of a laser radar (2.3.1.2) is greater than 70°, and the repeated scanning range of two adjacent laser radars (2.3.1.2) is 10%-30%; the shooting range (2.4.3) of the first camera (2.3.1.3) covers the scanning range of the corresponding laser radar (2.3.1.2); The S2 includes: S2.1, starting the driving element (2.1) of the detection module (2), so that the detection module (2) moves along the track (1), using the laser radar (2.3.1.2) to scan the tunnel lining surface (5) of the area to form a three-dimensional point cloud, and at the same time, the first camera (2.3.1.3) photographs the tunnel lining surface (5) of each area to obtain the tunnel image of each area; S2.

2. Construct a local coordinate system, and adjust the point cloud coordinates obtained by the laser radar scanning to the construction coordinates under the local coordinate system in real time based on the local coordinate system. The local coordinate system takes the center of the positioning module as the origin O, the vertical direction as the Z axis, the tunnel centerline direction as the Y axis, and the straight line direction perpendicular to the YOZ plane and passing through the origin as the X axis. Specifically: S2.2.

1. The emission point coordinates (a0, b0, c0) of the laser ranging element (3.3) are calculated based on the fixed installation distance, and the local coordinates (a1, b1, c1) of the laser landing point are calculated based on the emission point coordinates. The specific calculation formula is as follows: Wherein: l1 is the distance between the grid target (2.4.1) and the positioning module (3) measured by the laser distance measuring element (3.3); α1 is the real-time azimuth measured by the first electronic compass (3.5); β1 is the inclination measured by the first electronic compass (3.5); α0 is the initial azimuth in the Y-axis direction measured by the first electronic compass (3.5); S2.2.

2. Use the second camera (2.4.2) to obtain an image of the grid target (2.4.1), wherein the image of the grid target (2.4.1) is composed of white, green and red; S2.2.3, calculate the actual difference between the laser point and the center point of the grid target (2.4.1); S2.2.

4. Calculate the coordinates (a2, b2, c2) of the center point on the grid target (2.4.1) based on the azimuth and inclination measured by the second electronic compass (2.5). The specific calculation formula is as follows: Where: Δj is the vertical pixel coordinate difference between the laser point and the center point of the grid target; Δk is the horizontal pixel coordinate difference between the laser point and the center point of the grid target; is the average value of the actual size of a single pixel; α2 is the azimuth angle measured by the second electronic compass (2.5), and β2 is the inclination angle measured by the second electronic compass (2.5); S2.2.

5. Calculate the coordinates of the laser radar (2.3.1.2) and the second camera (2.4.2) respectively according to the fixed distance between the laser radar (2.3.1.2) and the second camera (2.4.2), and convert the point cloud coordinates scanned by each laser radar (2.3.1.2) into the construction coordinates in the local coordinate system; S2.3, projecting the construction coordinates of the three-dimensional point cloud into the UV coordinate system of the first camera (2.3.1.3) according to the external parameters of the first camera (2.3.1.3), and obtaining the pixel value of each point in the three-dimensional point cloud at the corresponding position of the corresponding tunnel image; S2.

4. Assign pixel values ​​to each point to form a colored three-dimensional point cloud for each area.

5. The method for detecting cracks on the tunnel lining surface according to claim 4, characterized in that: The S3 includes: S3.1, using statistical filtering algorithm to remove discrete points in the three-dimensional point cloud; S3.

2. Use the ICP algorithm to register the color point clouds of each area to obtain the three-dimensional point cloud of the entire tunnel, and use the octree algorithm to traverse and retrieve the three-dimensional point cloud of the entire tunnel to remove duplicate points; S3.

3. Expand the 3D point cloud of the entire tunnel according to the tunnel section contour line according to the point cloud expansion algorithm to obtain a 2D point cloud, specifically: The coordinates (x′, y′, z′) of the straight line segment point cloud in the three-dimensional point cloud are converted to the coordinates (x, y, z) of the XOY plane. The specific calculation formula is as follows: Where: G is the rotation matrix, T is the translation matrix; The point cloud coordinates (x i ,y i ,z i ) According to the cylindrical formula formed by the arc, the point cloud coordinates (x i ′,y i ′,z i ′), the specific calculation formula is as follows: Where: θ i For any point (x i ,y i ,z i ) with the axis at the center of the circle and the angle between the node and the node plane formed by the axis at the center of the circle. The node is the intersection point (x1, y1, z1) of the arc segment with other arc segments and straight line segments; is the angle between the tunnel centerline and the XOZ plane; (x0, y0, z0) is the coordinate of the axis at the center of the circle in a certain tunnel cross section; S3.

4. Project the 2D point cloud onto the XOY plane to obtain the second lining plan of the entire tunnel, specifically: S3.4.

1. Design the tunnel plan in the XOY plane based on the tunnel centerline and the perimeter of the second lining of the section; S3.4.

2. Divide the tunnel plane into a grid of 1 mm × 1 mm; S3.4.3, set the ordinate in the two-dimensional point cloud to 0, and then project the point cloud onto the tunnel plane empty map; S3.4.

4. Count the pixel values ​​of the point cloud in each grid and take the average value as the pixel value of the grid; S3.4.

5. Each grid cell forms a pixel point, forming a second lining plan of the entire tunnel.

6. The method for detecting cracks on the tunnel lining surface according to claim 5, characterized in that: The S4 includes: S4.

1. First, use labelme software to process the lining crack images of other tunnels and produce crack label images of set size; then use the U-Net algorithm to train and learn the produced crack label images to establish a tunnel crack semantic segmentation model. S4.2, cropping the full tunnel second lining plan in S3.4.5 according to the set size input by U-Net to obtain a cropped compliance image, filling the insufficient size part with blank pixels, and then numbering the cropped compliance images one by one according to the distance from the starting point and from the left side of the tunnel to the right side of the tunnel; S4.3, input the cropped compliance image into the tunnel crack semantic segmentation model for recognition, and extract its mask to obtain the corresponding pixel coordinates; S4.

4. All the recognized images are stitched together according to the numbers to form a tunnel-wide crack recognition result map, and the pixel coordinates of each crack in the tunnel-wide crack recognition result map are calculated.

7. A computer-readable storage medium, characterized in that: Computer program instructions are stored thereon, and when the computer program instructions are executed by a processor, a method for detecting cracks on the surface of a tunnel lining as claimed in any one of claims 1 to 6 is implemented.

8. An electronic device, characterized in that: include: At least one processor, at least one memory and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, the method for detecting cracks on the surface of a tunnel lining according to any one of claims 1 to 6.

Images