A lightning vision super-fusion-based structure crack damage real-time measurement method

Abstract

The application discloses a kind of structure crack real-time determination method based on radar vision superfusion, the method is with structure detection robot as ontology, robot integrates together with visible light video camera, laser radar;Depth information of structure crack and the distance between robot and crack physical position can be measured using laser radar, and real-time distance and structure crack depth data and video image complete feature fusion matching, and combined with improved artificial intelligence algorithm detects and separates crack position and calculates specific crack area size information, to achieve the detection effect of dynamically obtaining crack position plane size data and depth data, to realize the radar vision superfusion of longitudinal data measured by transverse plane visual information and laser radar, the method improves the efficiency and precision of crack perception, and has important reference significance for structure surface crack detection.

Description

Technical Field

[0001] This invention relates to the field of crack detection technology, specifically to a method for real-time measurement of structural cracks based on radar-visual hyperfusion. Background Technology

[0002] Tunnels are often constructed for transportation, mining, or other reasons. Tunnel cracks are one of the common structural damage diseases. There are many reasons for the formation of cracks, including construction factors, geological reasons, and the properties of the concrete itself.

[0003] Although the formation of cracks will not affect the initial structural safety of the tunnel, if they are not dealt with in time and allowed to develop, they will inevitably affect the service life of the tunnel and may even cause disasters. Therefore, it is extremely important to detect cracks in a timely manner.

[0004] Currently, routine inspections within tunnels lack regularity and are mostly conducted manually. This method is labor-intensive and often results in delayed detection of defects, potentially leading to serious accidents. Existing methods often rely on single automated detection systems without the cooperation of the inspection terminal, resulting in low overall inspection efficiency and effectiveness. Furthermore, another key aspect of crack treatment is determining whether the crack is currently stable, i.e., whether it is further developing, which is crucial for efficient crack management. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a real-time structural crack detection method based on radar-visual hyperfusion, which improves the efficiency of crack detection in tunnels, enhances detection quality, and increases processing efficiency.

[0006] To achieve the above objectives, the present invention provides a real-time structural crack damage measurement method based on laser-visual hyperconvergence. The real-time structural crack damage measurement method is applied to a structural crack damage detection system, which includes a structural inspection robot. The structural inspection robot integrates a camera, a lidar, and one or more modules for storing or executing the real-time measurement method.

[0007] The real-time measurement method includes the following steps:

[0008] S100. Determination of crack location: The structural inspection robot moves to the location of the crack via a track, collects the location information of the crack, and records the time of crack discovery;

[0009] S200. Perform crack damage detection, including the following methods:

[0010] S201. Collect image data of the crack location through a camera, extract crack features from the image data, establish the correspondence between internal and external parameters based on the extracted features, segment the crack based on the extracted features, and output the crack segmentation image.

[0011] The spatial location and depth information of the cracks are collected using lidar; the spatial location and depth feature information of the cracks are extracted.

[0012] S202. By associating the radar target and image icon with the spatial location and depth feature information of the crack through the correspondence between the internal and external parameters, the radar-visual information is fused, and a radar-visual hyper-fusion image is output.

[0013] S203. Perform a secondary fusion of the crack segmentation image and the radar-visual hyper-fusion image to obtain a real crack image with real-time crack parameter information;

[0014] S204. Compare the real-time parameter information with the crack level classification standard and output the real-time measurement results.

[0015] Based on the above-mentioned method for determining the degree of crack damage, the crack detection and segmentation model trained using the image segmentation algorithm can efficiently and quickly detect crack damage. After detecting crack damage, the structural inspection robot slides along the track and reaches the crack location. During the process of reaching the crack, the location information of the crack is recorded and saved. The depth of the structural crack and the distance between the robot and the physical location of the crack can be measured using LiDAR. The distance and spatial depth feature data are fused with the two-dimensional feature information of the video image in real time.

[0016] According to the present invention, a method for real-time structural crack damage determination based on radar-visual hyperfusion is provided, which involves secondary fusion of crack segmentation images and radar-visual hyperfusion images, including: projecting the features of the crack segmentation images into 3D space to achieve image feature fusion; and fusing the crack segmentation images and radar-visual hyperfusion images of different resolutions together by using continuous convolution to obtain the real crack image.

[0017] According to the present invention, a real-time structural crack damage measurement method based on laser radar hyperfusion is provided, wherein the crack segmentation image and the laser radar hyperfusion image at different resolutions are fused together by using continuous convolution, including: obtaining a profile feature map of the crack segmentation image; encoding the 3D adjacent offset between the laser radar point and the target pixel on the profile feature map; and extracting information from the profile feature image closest to each laser radar point through encoding, thereby completing the fusion of multiple features.

[0018] According to the present invention, a method for real-time structural crack damage determination based on radar-visual hyperfusion is provided. The method involves segmenting cracks by extracting features and outputting crack segmentation images, including: judging the segmented images and outputting the judgment result; if the judgment result is an independent long thin strip crack, the output crack segmentation image contains the maximum length and maximum width information of the crack; if the judgment result is a non-independent long thin strip crack, the output crack segmentation image contains the area information of the crack.

[0019] In this way, a method for measuring the degree of cracking that integrates two-dimensional image features and spatial depth features is deployed on a dynamic tunnel inspection robot. Combined with an improved artificial intelligence algorithm, the crack location is detected, segmented, and the planar crack area is calculated, while the depth data of the cracked area is dynamically acquired in real time. This enables multi-dimensional real-time measurement of the structural cracking condition. This method improves the efficiency and accuracy of crack perception and has important reference value for the detection of cracks on structural surfaces.

[0020] According to the present invention, a method for real-time structural crack damage determination based on radar-visual hyperconvergence is provided. The method for comparing the real-time parameter information with the crack level classification standard and outputting the real-time measurement result includes: setting the judgment thresholds for independent fine strip cracks and non-independent fine strip cracks respectively; comparing the detected crack information with the judgment thresholds and outputting three types of crack level judgment results: minor crack damage, ordinary crack damage, and severe crack damage.

[0021] According to the present invention, a method for real-time determination of structural crack damage based on radar-visual hyperconvergence is provided. The structural inspection robot moves to the location of the crack by a track and collects the location information of the crack, including: obtaining the actual specific location of the disaster in the structure by the number of gear rolls of the structural inspection robot moving on the track, the correspondence between the RFID on the track and the station number, and the reaction of the magnetic strip on the robot.

[0022] In this way, when determining the location, the robot's geographical coordinates can be determined by correlating the RFID tags on the track with the station numbers and the magnetic stripe on the robot. This means that when the robot is deployed in multiple tunnels, this method can effectively pinpoint which tunnel the specific disaster location is in. Then, by using the number of gear rotations and the RFID tags in conjunction, the robot's exact position within the tunnel can be determined, thus providing a precise understanding of the robot's location within the tunnel. Therefore, compared to existing devices, this method of disaster location determination is more accurate, facilitating the development of solutions tailored to the specific external environment.

[0023] According to the present invention, a method for real-time determination of structural crack damage based on radar-visual hyperfusion is provided, wherein the radar-visual information is fused by associating the internal and external parameter calibration correspondence with the crack spatial location and depth feature information with radar targets and image icons, and outputting a radar-visual hyperfusion image, including: projecting the point cloud data depth map onto the image plane by frame-by-frame tracking fusion to complete feature matching;

[0024] Using photometric loss, dense pixel errors between the predicted depth map and the correct depth map are checked; using point cloud distance loss to reflect the point cloud transformation allowed by the 3D spatial transformer layer after backprojection, an attempt is made to minimize the 3D-3D point distance between the uncalibrated transformation point and the target point cloud; combining the above scene association loss function, global features are aggregated, and regression is completed for error calibration, outputting the corrected and calibrated Ravage hyperfusion image.

[0025] According to the present invention, a method for real-time determination of structural crack damage based on radar-visual hyperfusion is provided. If the determination result is an independent long and thin strip crack, the output crack segmentation image contains the maximum length information and the maximum width information of the crack, including: setting the ratio parameter b of pixels to actual length based on the specific physical distance at the time of shooting; calculating the pixel value of the segmented crack, extracting the maximum pixel width d and pixel length l in the crack, and calculating the maximum width D = b*d and the maximum length L = b*l of the crack.

[0026] According to the present invention, a method for real-time determination of structural crack damage based on radar-visual hyperfusion is provided. If the structure is determined to be a non-independent long and thin strip crack, the output crack segmentation image contains crack area information, including: setting a ratio parameter b between pixels and actual length based on the specific physical distance at the time of shooting; using the ContourArea function in the OpenCV library to calculate the pixel area S1 within the contour using Green's formula, and obtaining the actual crack damage area S = b * b * S1.

[0027] The beneficial effects of this invention are as follows: By establishing and training a crack detection model, crack detection becomes more efficient. A structural inspection robot is activated to specifically detect the location and condition of cracks. Combined with LiDAR, crack depth is measured. Features captured by both visual and LiDAR modules are fused to achieve hyper-fusion of lateral planar visual information and longitudinal data measured by LiDAR, improving the accuracy of object detection and recognition. This hyper-fusion provides a more specific understanding of the extent of tunnel cracking, enabling more efficient and accurate development of emergency response plans. Furthermore, the marking and periodic re-inspection of small cracks improves the efficiency and quality of crack handling within the tunnel. Attached Figure Description

[0028] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0029] Figure 1 This is a partial structural block diagram of a structural inspection robot provided in an embodiment of the present invention;

[0030] Figure 2 This is a flowchart of the present invention;

[0031] Figure 3 for Figure 2 A flowchart detailing the parsing steps;

[0032] Figure 4 for Figure 2 A flowchart detailing the parsing steps;

[0033] Figure 5 for Figure 2 A flowchart detailing the parsing steps;

[0034] Figure 6 for Figure 2 A flowchart detailing the parsing steps;

[0035] Figure 7 This is an actual test of another embodiment of the present invention.

[0036] In the attached diagram, the components include a structural inspection robot 100, a data processor 101, a memory 102, a storage controller 103, a peripheral interface 104, a lidar 105, and a power supply 107. Detailed Implementation

[0037] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0038] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art, and numerous specific details are set forth in the following detailed description for the purpose of fully understanding the invention. However, those skilled in the art should understand that the invention can be implemented without these specific details.

[0039] Figure 1The present invention describes a real-time structural crack damage measurement method based on laser-visual hyperconvergence, which is applied to a crack detection system in a tunnel. The measured cracks are structural cracks in the tunnel; therefore, the following embodiments all use tunnel cracks as the specific actual measurement system. The structural crack damage detection system includes a structural inspection robot 100, which includes a sliding connection structure that can be attached to and slide along a track. The robot 100 includes a drive motor, a camera, a memory 102, a memory controller 103, one or more data processors 101, a peripheral interface 104, a lidar 105, and one or more modules for storing or executing automatic detection methods. It also includes a battery pack or other power supply port that provides power 107 to the above components. These components are communicatively connected to a server or control terminal via one or more communication buses, signal lines, or wireless communication modules. The control terminal or server has a display screen that can display the detection information. It should be understood that the structural inspection robot 100 is merely one instance of this application, and the terminal device may have more or fewer components than illustrated. Each component can be implemented using hardware, software, or a combination of both, including one or more data processors 101 and / or application-specific integrated circuits, and of course, a power system that provides power to the structural inspection robot 100. Upon detecting an abnormal crack, the structural inspection robot 100 will execute an automatic detection method to detect the crack and issue an early warning through an early warning system.

[0040] The drive motors include motors that generally convert electrical energy into driving torque, as well as servo motors. In some embodiments, servo motors are used to detect and collect the position of the structure inspection robot 100.

[0041] The memory 102 controller can control access to the memory 102 by devices such as the data processor 101 in the structure inspection robot 100. The memory 102 contains multiple software programs and / or instruction sets, and can connect inputs or outputs to the memory 102 or the processor through the peripheral interface 104. This enables the structure inspection robot 100 to perform various inspection functions and process data. In some embodiments, the memory 102 stores various inspection models trained by image segmentation algorithms, such as crack detection models. These models can be used to detect various disaster situations in tunnels. The structure inspection robot 100 can run these models and implement the inspection method. The peripheral interface 104 connects to external inspection devices, such as lidar and high-definition cameras, for detecting the external environment and transmitting data.

[0042] The automatic detection method includes the following steps:

[0043] In step S100, the location of the crack is determined by the structural inspection robot moving to the location of the crack via a track, collecting the location information of the crack, and recording the time of crack discovery.

[0044] In step S200, crack damage detection is performed, and the detection method includes:

[0045] S201: Image data of the crack location is acquired using a camera; crack features are extracted from the image data; an internal and external parameter calibration correspondence is established based on the extracted features; crack segmentation is performed using the extracted features; and a crack segmentation image is output. Spatial location and depth information of the crack are acquired using a LiDAR; and spatial location and depth feature information of the crack are extracted. In a specific embodiment, a convolutional neural network is used to extract features from the cracked areas in the acquired image data and perform detection and segmentation, outputting a crack segmentation image.

[0046] In the above embodiments, the spatial location and depth information of the crack are determined by the lidar reflection range and wavelength function using formula one.

[0047]

[0048] Formula (1) describes how the radiance of an optical system changes with distance and wavelength. The physical distance of the robot to the detected defect is determined using the principle of this formula; where L on the left side of Formula (1)... r (ρ, λ) represents a physical quantity related to an optical system, commonly known as radiance, measured in watts per square meter per solid radian; ρ on the right side of Equation 1 represents distance, measured in meters (m); λ represents wavelength, measured in nanometers (nm); and I0 represents incident light intensity, measured in watts per square meter (W / m²). -2 η represents the optical efficiency of the receiving system, which is usually a number less than 1; A represents the area illuminated by the incident light, in square meters (m²). 2 β(ρ, λ) represents the absorption coefficient of the medium, which is a function of distance and wavelength, and its unit is decibels per meter (dBm). -1 σ(r,λ) represents the scattering coefficient of the medium, which is a function of distance and wavelength, and its unit is decibels per meter (dBm). -1 The exponential term in the above formula; This represents the scattering attenuation of the medium, where the integral is twice the sum of the scattering coefficients over all distances from the incident point to the observation point in the medium.

[0049] S202, by associating the radar target and image icon with the spatial location and depth feature information of the crack through the correspondence between the internal and external parameters calibration, radar-visual information fusion is performed, and a radar-visual hyper-fusion image is output.

[0050] Specifically, the point cloud data depth map is projected onto the image plane through frame-by-frame tracking fusion to complete feature matching; photometric loss is used to check the dense pixel error between the predicted depth map and the correct depth map, as shown in Formula 2:

[0051]

[0052] In formula (2), the value used to measure the difference between the predicted and true values ​​is typically the optimization objective used in training machine learning models such as neural networks. This formula represents a mean squared error loss function, which is half the sum of the squares of the differences between the predicted and true values. Therefore, the smaller this loss function is, the smaller the difference between the predicted and true values. Where N represents the number of samples in the dataset; D... gt π represents the ground truth, typically indicating the true label or target value of a sample in the dataset; K represents a scaling factor, usually a scalar; T represents a transformation matrix, usually a two-dimensional matrix; π represents pi, approximately 3.14; D miscalib This represents a miscalibration value, which is usually a parameter related to equipment calibration.

[0053] The point cloud distance loss function, which reflects the 3D spatial transformer layer's ability to transform the point cloud after backprojection, attempts to minimize the 3D-3D point distance between the uncalibrated transformed points and the target point cloud. Combined with the scene association loss function, global features are aggregated, and regression is performed to correct for errors in calibration, outputting the corrected and calibrated radar-visual hyper-fusion image. The entire loss function is defined as Equation 3:

[0054]

[0055] In formula (3), d icp (s1,s2) represents the ICP error metric between two point clouds s1 and s2. ICP (Iterative Closest Point) is an algorithm for point cloud registration, used to find the rigid transformations (rotation matrix and translation vector) between two point clouds to best align them; N represents the number of points in the point cloud; i represents the i-th point. Represents the coordinates of the i-th point in the reference point cloud; (RX)miscalib +t) represents the coordinates of the i-th point in the source point cloud after rotation R and translation t, where R and t are unknown rigid body transformation parameters; |||| represents the Euclidean distance; X miscalib This represents a miscalibration value, which is usually a parameter related to equipment calibration and used to compensate for equipment errors.

[0056] S203, the crack segmentation image and the radar-visual hyper-fusion image are fused a second time to obtain a real crack image with real-time crack parameter information.

[0057] Specifically, the crack segmentation image features are projected into 3D space to achieve image feature fusion; the crack segmentation images and the radar hyper-fusion image at different resolutions are fused together using continuous convolution to obtain the real crack image. In a further embodiment, a profile feature map of the crack segmentation image is obtained; the 3D adjacent offset between the lidar point and the target pixel on the profile feature map is encoded; information is extracted from the profile feature image closest to each lidar point through encoding, thereby completing the fusion of multiple features.

[0058] S204, compare the real-time parameter information with the crack level classification standard and output the real-time measurement results.

[0059] Specifically, the segmented image is judged, and the judgment result is output; if the judgment result is an independent long thin strip crack, then the output crack segmentation image contains the maximum length and maximum width information of the crack. The ContourArea function in the OpenCV library is used to calculate the pixel area S1 within the contour using Green's formula, Equation 4:

[0060]

[0061] Green's formula (4) represents the circulation or line integral on curve L. It represents the result of integrating Pdx and Qdy on curve L, where P and Q are vector fields on the plane of L, and dx and dy represent small path elements; the right side of the formula... Let represent the double integral of region D, which represents the result of taking the divergence of the vector field within D, where and Let represent the partial derivatives of vector fields Q and P in the x and y directions, respectively.

[0062] Based on the specific physical distance during shooting, the ratio parameter b between pixels and actual length is set to obtain the actual crack area S = b * b * S1.

[0063] Furthermore, if the structure is determined to be a non-independent long, thin strip crack, the output crack segmentation image contains crack area information. Pixel values ​​are calculated for the segmented cracks, and the maximum pixel width d and pixel length l in the crack are extracted. Using the ratio parameter b between the set pixel and actual length, the maximum crack width D = b * d and the maximum length L = b * l are calculated.

[0064] In a further embodiment, the cracks are classified into different levels, including setting threshold values ​​for independent fine-strip cracks and non-independent fine-strip cracks.

[0065] The detected crack information is compared with the judgment threshold, and the crack level judgment results of three categories, namely, minor crack damage, ordinary crack damage, and severe crack damage, are output.

[0066] Specifically, the system for structural inspection robots sets judgment criteria, initially defining three types of cracks: independent slender cracks, independent large cross-sectional area cracks, and large-area crack clusters. Severity evaluation thresholds are set for each type of crack. Regional values ​​and type statistics are statistically analyzed for the crack size data and verification degree classification obtained from each measurement. Distribution analysis of the size values ​​of different types of cracks appearing in the tunnel is performed. Based on the analysis results, the judgment criteria set by the system are continuously updated. This ensures that the output real-time measurement results integrate crack area, crack depth, and width data, and complete the determination and judgment of the crack level at the crack location.

[0067] In the above embodiment, after uploading data for cracks that have not reached a predetermined threshold, the system marks the location, so that the structural inspection robot 100 repeatedly inspects the marked cracks within a predetermined time period. Multiple repeated inspection results are compared sequentially according to the inspection time. The comparison results determine whether the crack has developed. If the crack has developed, the inspection time period is shortened and the data is regenerated and uploaded; if the crack is stable, the inspection time period is delayed.

[0068] In one specific embodiment, the actual location of the disaster in the tunnel is obtained by detecting the number of gear rolls of the structure detection robot 100 moving on the track, the correspondence between the RFID on the track and the station number, and the reaction of the magnetic strip on the robot.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for real-time structural crack damage measurement based on radar-visual hyperconvergence, wherein the method is applied in a structural crack damage detection system, the structural crack damage detection system comprising a structural inspection robot, characterized in that: The structural inspection robot integrates a camera, a lidar, and one or more modules for storing or executing real-time measurement methods. The real-time measurement method includes the following steps: S100. Determination of crack location: The structural inspection robot moves to the location of the crack via a track, collects the location information of the crack, and records the time of crack discovery; S200. Perform crack damage detection, including the following methods: S201. Collect image data of the crack location through a camera, extract crack features from the image data, establish the correspondence between internal and external parameters based on the extracted features, segment the crack based on the extracted features, and output the crack segmentation image. The spatial location and depth information of the cracks are collected using lidar; the spatial location and depth feature information of the cracks are extracted. S202. By associating the radar target and image icon with the spatial location and depth feature information of the crack through the correspondence between the internal and external parameters, the radar-visual information is fused, and a radar-visual hyper-fusion image is output. S203. Perform a secondary fusion of the crack segmentation image and the radar-visual hyper-fusion image to obtain a real crack image with real-time crack parameter information; S204. Compare the real-time parameter information with the crack grade classification standard and output the real-time measurement results; The step of performing a secondary fusion of the crack segmentation image and the radar-visual hyper-fusion image includes: The crack segmentation image features are projected into 3D space to achieve image feature fusion; The real crack image is obtained by fusing the crack segmentation images and the radar hyper-fusion images at different resolutions together using continuous convolution; The process of fusing the segmented crack images and the radar-visualized hyper-fusion images at different resolutions using sequential convolution includes: Obtain the cross-sectional feature map of the crack segmentation image; Encode the 3D adjacent offsets between the lidar points and the target pixels on the profile feature map; Information is extracted from the profile feature image closest to each LiDAR point by encoding, thereby completing the fusion of multiple features.

2. The method for real-time measurement of structural crack damage according to claim 1, characterized in that, The step of segmenting cracks using extracted features and outputting a crack segmentation image includes: The segmented image is evaluated, and the evaluation result is output. If the judgment result is an independent long and thin strip crack, the output crack segmentation image contains the maximum crack length information and the maximum crack width information; If the structure is determined to be a non-independent long and thin strip crack, then the output crack segmentation image contains crack area information.

3. The method for real-time measurement of structural crack damage according to claim 2, characterized in that, The step of comparing the real-time parameter information with the crack grade classification standard and outputting the real-time measurement result includes: Set separate thresholds for determining independent fine-grained cracks and non-independent fine-grained cracks; The detected crack information is compared with the judgment threshold, and the crack level judgment results of three categories, namely, minor crack damage, ordinary crack damage, and severe crack damage, are output.

4. The method for real-time measurement of structural crack damage according to claim 1, characterized in that, The structural inspection robot moves along a track to the location of the crack and collects the location information of the crack, including: the number of gear rotations during the robot's movement on the track, the correspondence between the RFID tag on the track and the station number, and the reaction of the magnetic strip on the robot, to obtain the actual specific location of the disaster in the structure.

5. The method for real-time measurement of structural crack damage according to claim 1, characterized in that, The process of associating radar targets and image icons with the spatial location and depth feature information of the crack through the correspondence between the internal and external parameters to perform radar-visual information fusion and outputting a radar-visual hyper-fusion image includes: The point cloud data depth map is projected onto the image plane by frame-by-frame tracking and fusion to complete feature matching. Using photometric loss, we examine dense pixel errors between the predicted depth map and the correct depth map; By utilizing the point cloud distance loss response 3D spatial transformer layer to allow point cloud transformation after backprojection, we attempt to minimize the 3D-3D point distance between the uncalibrated transformed point and the target point cloud using a metric scale. By combining the scene association loss function, aggregating global features, and then performing regression for error calibration, the corrected and calibrated radar-visual hyper-fusion image is output.

6. The method for real-time measurement of structural crack damage according to claim 2, characterized in that, If the determination result is an independent, long, thin strip-shaped crack, then the output crack segmentation image contains information on the maximum crack length and maximum crack width, including: Based on the specific physical distance during shooting, set the ratio parameter b between pixels and actual length; Pixel values ​​are calculated for the segmented cracks, and the maximum pixel width d and pixel length l in the cracks are extracted. The maximum width D = b * d and the maximum length L = b * l of the cracks are then calculated.

7. The method for real-time measurement of structural crack damage according to claim 2, characterized in that, If the structure is determined to be a non-independent long, thin strip crack, the output crack segmentation image contains crack area information, including: Based on the specific physical distance during shooting, set the ratio parameter b between pixels and actual length; The ContourArea function in the OpenCV library is used to calculate the pixel area S1 within the contour using Green's formula, thus obtaining the actual crack area S = b * b * S1.

Images