Intelligent Measurement Method and System for Structural Cracks Based on Laser Measurement

Abstract

This invention discloses an intelligent measurement method and system for structural cracks based on laser measurement, belonging to the field of structural inspection and measurement technology. The method includes acquiring a high-density three-dimensional point cloud of the structural surface through laser scanning, segmenting the point cloud of candidate crack regions and the point cloud of the complete structural surface. Geometric morphological refinement and skeleton extraction are performed on the candidate point cloud to generate crack skeleton lines and construct its topological connection map, thereby accurately calculating the crack length, width, and orientation angle. By fusing the crack geometric parameters with the structural surface reference information, the spatial position parameters of the crack on the structure are calculated. This method ensures the accuracy of calculating the geometric parameters of complex cracks through skeleton extraction and topological analysis, and provides a precise spatial orientation description of the crack by fusing structural reference information, ultimately generating a comprehensive measurement report including geometric morphology and spatial position.

Description

Technical Field

[0001] This invention belongs to the field of structural inspection and measurement technology, specifically a method and system for intelligent measurement of structural cracks based on laser measurement. Background Technology

[0002] In the field of structural health monitoring and safety assessment, accurate measurement of surface cracks in concrete, steel structures, and other materials is a crucial step. Existing crack measurement technologies primarily rely on digital image processing methods. These methods acquire two-dimensional images of the structural surface and use image processing algorithms to identify and quantify crack features. However, two-dimensional image methods are susceptible to interference from ambient lighting conditions, shadows, and surface contaminants, resulting in insufficient measurement accuracy and stability, and they cannot directly obtain the three-dimensional geometric information of the cracks. Another contact-based measurement method is inefficient and may cause secondary damage to fragile structures.

[0003] Existing technologies have limitations in acquiring the true three-dimensional morphology of cracks and their spatial relationship with the overall structure. Image-based methods can only provide two-dimensional projection information, failing to accurately reflect the true direction, depth trend, and spatial distribution of cracks on the structural surface. Measurement results are limited to the apparent width and length of the cracks themselves, lacking the ability to correlate crack dimensions with the geometric coordinate system of the structure. This results in measurement reports with limited information dimensions, making them difficult to directly use for assessing the specific impact of cracks on the overall structural performance.

[0004] A technical solution is needed that can directly capture the complete geometry of cracks in three-dimensional space and accurately determine the specific location of cracks on the surface of complex structures. This solution must overcome the limitations of two-dimensional measurement, solve the problems of centerline extraction and topological description of complex crack networks, and achieve automatic correlation and calculation of crack geometric parameters with the spatial reference of the structural entity. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art; Therefore, this invention proposes a smart measurement method for structural cracks based on laser measurement, comprising: A laser measurement instrument is used to perform regional scanning measurement on the surface of the target structure to obtain a high-density three-dimensional point cloud dataset containing crack information. Perform surface structure feature segmentation processing on the high-density three-dimensional point cloud dataset to separate the background point cloud subset corresponding to the complete structure surface and the candidate point cloud subset corresponding to the suspected crack region; Geometric morphological thinning and skeleton extraction are performed on the candidate point cloud subset to generate crack skeleton line point cloud, and topological connectivity analysis is performed on the crack skeleton line point cloud to construct the crack topological connectivity graph. Based on the crack skeleton line point cloud and the topological connection map, the set of geometric morphological parameters of the crack is calculated, including crack length, average width, maximum width and orientation angle. The set of geometric morphology parameters is fused with the structural surface reference information extracted from the background point cloud subset to calculate the spatial position parameters of the crack relative to the structural surface, generating a comprehensive measurement report that includes the crack's geometry and spatial position.

[0006] Further, surface structure feature segmentation processing is performed on the high-density 3D point cloud dataset to separate the background point cloud subset corresponding to the intact structure surface and the candidate point cloud subset corresponding to the suspected crack region, including: Calculate the local curvature features and normal vector of each point in the high-density 3D point cloud dataset, and identify potential discontinuities based on the degree of abrupt change in the local curvature features and normal vector; For each potential discontinuity point, a local neighborhood point set is constructed, and the eigenvalues ​​of the three-dimensional spatial distribution covariance matrix of the local neighborhood point set are calculated. Based on the ratio relationship of the aforementioned feature values, linear distribution features and planar distribution features are distinguished, and local neighborhood point sets exhibiting obvious linear distribution features are initially marked as crack candidate points. A region growing operation is performed on the initially marked crack candidate points to merge spatially adjacent crack candidate points with the same normal direction, forming a continuous set of suspected crack region point clouds, i.e., the candidate point cloud subset. The candidate point cloud subset is removed from the original high-density 3D point cloud dataset, and the remaining point clouds constitute the background point cloud subset.

[0007] Further, geometric morphological thinning and skeleton extraction processing are performed on the candidate point cloud subset to generate a crack skeleton line point cloud, including: The candidate point cloud subset is subjected to voxel rasterization downsampling processing to generate a regularized three-dimensional voxel mesh model; In the three-dimensional voxel mesh model, an iterative erosion operation is performed on the voxels marked as crack regions until the remaining voxels form a linear structure with a single voxel width. The remaining voxel center points after iterative erosion are extracted to form the initial skeleton point set; The initial skeleton point set is subjected to smoothing filtering and redundant point removal to ensure the continuity of the skeleton lines and the width of a single pixel, thereby generating a refined crack skeleton line point cloud.

[0008] Furthermore, a topological connectivity analysis is performed on the point cloud of the crack skeleton lines to construct a topological connectivity graph of the cracks, including: Key points are identified in the point cloud of the crack skeleton line, and the key points include the endpoints, branch points and intersections of the skeleton line. Calculate the three-dimensional Euclidean distance between all point pairs in the crack skeleton line point cloud, and determine the direct connection relationship between points based on a preset distance threshold; Using the identified key points as nodes and the paths formed by continuous skeleton points between key points as edges, an undirected graph structure is constructed. Redundant edge elimination and ring structure parsing are performed on the undirected graph structure to ensure that the undirected graph structure can accurately reflect the real topological connection relationship of the crack network, and the final crack topological connection graph is generated.

[0009] Furthermore, based on the crack skeleton line point cloud and the topological connectivity graph, the set of geometric morphological parameters of the crack is calculated, including: In the topological connection graph, each edge connecting two endpoints is traversed, and the sum of the line segment lengths between all consecutive points on the skeleton path corresponding to the edge is calculated as the length of the crack branch. For each skeleton point in the crack skeleton line point cloud, in its normal plane, start from the skeleton point and search along both sides along the normal direction until the boundary of the background point cloud subset is reached, and the sum of the search distances on both sides is taken as the local width of the crack at the skeleton point. The local width of the crack at all skeleton points on a crack branch is statistically analyzed, and the average value is calculated as the average width of the crack branch. The maximum value among them is taken as the maximum width of the crack branch. Calculate the principal direction vector of the crack branch skeleton line, and calculate the angle between the principal direction vector and the structural surface reference plane extracted from the background point cloud subset, as the direction angle of the crack branch; The length, average width, maximum width, and orientation angle of all crack branches are summarized to form a set of geometric parameters for the crack.

[0010] Furthermore, the structural surface reference information extracted from the background point cloud subset includes: A plane fitting process based on random sampling consistency is performed on the background point cloud subset to extract the principal structure plane equation of the target structure; Calculate the normal vector of the main structural plane in three-dimensional space and the range of the plane boundary; Using the main structural plane as a reference, a local two-dimensional coordinate system is established, and the points in the background point cloud subset are projected onto the local two-dimensional coordinate system to obtain a two-dimensional projection point set of the structural surface; Based on the two-dimensional projection point set, the flatness deviation and local undulation characteristics of the structural surface are calculated as a supplement to the reference information of the structural surface.

[0011] Further, the set of geometric morphological parameters is fused with the structural surface reference information extracted from the background point cloud subset to calculate the spatial position parameters of the crack relative to the structural surface, including: Project each point in the crack skeleton line point cloud onto the main structural plane to obtain the two-dimensional projection trajectory of the crack on the structural surface. In the local two-dimensional coordinate system, the centroid coordinates of the two-dimensional projection trajectory of the crack are calculated as the planar position of the crack on the surface of the structure. Calculate the average distance from all points in the crack skeleton line point cloud to the main structural plane, and use it as the average embedment depth of the crack relative to the structural surface; Calculate the angle between the crack direction angle and the main direction of local undulations on the structural surface, and use it as the relative angle between the crack direction and the surface texture. The spatial position parameters of the crack are constituted by the planar position of the crack, the average embedment depth, and the relative angle between the crack direction and the surface texture.

[0012] Furthermore, the generation of a comprehensive measurement report including the crack geometry and spatial location includes: Establish a standardized crack measurement report data structure template, which includes fields for storing crack identifiers, geometric parameters, spatial location parameters, and measurement timestamps; The calculated set of geometric morphological parameters of the crack and the spatial location parameters of the crack are filled in according to the field format of the data structure template; The crack topology connection diagram is encoded in vector graphics format and associated with the corresponding crack identifier; All the filled and coded data are integrated to generate a structured data file, namely the comprehensive measurement report.

[0013] Furthermore, before performing area scanning measurement on the target structure surface using a laser measuring instrument, a measurement planning step is also included: Obtain design drawings or a rough 3D model of the target structure and identify the crack-prone areas that require key monitoring; Based on the spatial distribution of the easily cracked areas, the scanning path and station layout of the laser measuring instrument are planned to ensure complete coverage of the target area. Based on the surface material of the target structure and the expected crack size, set the scanning resolution and sampling frequency parameters of the laser measuring instrument; Based on the planned scanning path and site layout, the laser measuring instrument is controlled to automatically perform the area scanning measurement task.

[0014] Furthermore, the present invention also includes a laser-based intelligent structural crack measurement system, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the steps of the laser-based intelligent structural crack measurement method described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By performing geometric morphological refinement and skeleton extraction on a subset of candidate point clouds and constructing a topological connectivity graph, the central skeleton line of a crack can be directly extracted from three-dimensional space, and its topological connectivity can be accurately described. This scheme solves the fundamental problem of mathematically representing complex crack morphologies. Crack length is calculated based on the central skeleton line, and its measurement path strictly follows the actual extension trajectory of the crack on the three-dimensional surface, eliminating path deviations caused by traditional two-dimensional projection or local point sampling methods. The explicit topological connectivity provides a rigorous logical basis for calculating crack direction and identifying branches and intersections, avoiding misjudgments of topological structure caused by point cloud noise or irregular morphology, thus achieving stable and reliable quantification of the basic geometric parameters of cracks of arbitrary shapes.

[0016] By fusing the set of geometric parameters of the cracks with structural surface reference information extracted from a subset of the background point cloud, a unified spatial coordinate system for crack parameters and the structural body is achieved. This scheme establishes a spatial relationship between the geometric dimensions of the cracks and the structural entity to which they belong. The calculated spatial location parameters of the cracks can directly describe the specific orientation of the cracks on the structural surface, as well as their distance and angle relative to the design reference. This transforms the measurement results from isolated dimensional data into information with engineering spatial semantics, which can be directly used to analyze the potential relationship between cracks and structural stress distribution and construction details, providing a quantitative basis containing precise location context for structural condition assessment. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of the intelligent structural crack measurement method based on laser measurement described in this invention. Figure 2 A flowchart for geometric morphological refinement and skeleton extraction of candidate point cloud subsets; Figure 3 A flowchart for the topological connectivity analysis of the point cloud of the crack skeleton line; Figure 4 Box plots showing the crack width distribution for three target types; Figure 5 A biaxial analysis chart showing the risk of crack-prone areas of bridges and the statistics of historical cracks. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See Figure 1 A high-density 3D point cloud dataset containing crack information was obtained by scanning the surface of the target structure using a laser measuring instrument. Surface structure feature segmentation was performed on the high-density 3D point cloud dataset to separate a background point cloud subset corresponding to the intact structural surface and a candidate point cloud subset corresponding to suspected crack areas. Geometric morphological refinement and skeleton extraction were performed on the candidate point cloud subset to generate crack skeleton line point clouds, and topological connectivity analysis was performed on the crack skeleton line point clouds to construct a crack topological connectivity map. Based on the crack skeleton line point clouds and the topological connectivity map, a set of geometric morphological parameters of the crack was calculated, including crack length, average width, maximum width, and orientation angle. The set of geometric morphological parameters was fused with the structural surface reference information extracted from the background point cloud subset to calculate the spatial position parameters of the crack relative to the structural surface, generating a comprehensive measurement report containing the crack's geometry and spatial position.

[0020] See Figure 2 In one embodiment of the present invention, the local curvature features and normal vectors of each point in a high-density three-dimensional point cloud dataset are calculated, and potential discontinuities are identified based on the degree of abrupt changes in the local curvature features and normal vectors. In a specific implementation, for a scanned point cloud of a concrete beam surface, principal component analysis is used to calculate the covariance matrix of each point and its radius neighborhood. The eigenvector corresponding to the minimum eigenvalue of the covariance matrix is ​​the estimate of the normal vector of that point. The maximum and minimum principal curvatures of the point are calculated, and the standard deviation of the curvature values ​​of each point in the neighborhood is compared with a preset curvature abrupt change threshold. When the standard deviation exceeds the threshold, the point is marked as a potential discontinuity. A local neighborhood point set is constructed centered on each potential discontinuity, and the eigenvalues ​​of the three-dimensional spatial distribution covariance matrix of the local neighborhood point set are calculated. In some embodiments, for a potential discontinuity, all points within a spherical neighborhood of radius R are selected to form a local neighborhood point set. The covariance matrix of this point set is calculated and eigenvalue decomposition is performed to obtain three eigenvalues. and satisfy Based on the ratio relationship of eigenvalues, linear and planar distribution features are distinguished, and local neighborhood point sets exhibiting obvious linear distribution features are initially marked as crack candidate points. Optionally, a linear feature metric can be used for determination, defined by the following formula:

[0021] Where: symbol The calculation result of the linear characteristic metric, with the symbol... The largest eigenvalue of the three-dimensional spatial distribution covariance matrix is ​​represented by the symbol. This represents the intermediate eigenvalues ​​of the three-dimensional spatial distribution covariance matrix. When... When the value is greater than a preset linear feature threshold, the local neighborhood point set is determined to exhibit obvious linear distribution characteristics, and all points contained therein are initially marked as crack candidate points. A region growing operation is performed on the initially marked crack candidate points to merge spatially adjacent crack candidate points with consistent normals, forming a continuous set of suspected crack region point clouds, i.e., a candidate point cloud subset. It can be understood that the seed point for the region growing operation is selected from all unvisited crack candidate points. The growth criteria include that the 3D Euclidean distance between points is less than a distance threshold, and the angle between the point normal vectors is less than an angle threshold. Points satisfying the growth criteria are merged into the same region. The candidate point cloud subset is removed from the original high-density 3D point cloud dataset, and the remaining point cloud constitutes the background point cloud subset.

[0022] A voxel rasterization downsampling process is performed on a subset of candidate point clouds to generate a regularized 3D voxel mesh model. In the specific implementation, a voxel size is set, and the 3D space is divided into a regular cubic grid. Within each voxel grid, the position of the voxel is represented by the centroid coordinates of all candidate point cloud subset points within it. If a voxel grid contains at least one point, the voxel is marked as "active," indicating that it belongs to a crack region. In the 3D voxel mesh model, iterative erosion operations are performed on the voxels marked as crack regions until the remaining voxels form a linear structure the width of a single voxel. In some embodiments, the iterative erosion operation uses a six-neighbor or twenty-six-neighbor erosion template in 3D morphology. In each iteration, all neighboring voxels of each active voxel are checked. If any neighboring voxel is in an "inactive" state, the active voxel is marked as "to be removed" in this iteration. After all active voxels have been checked in this iteration, all voxels marked as "to be removed" are set to "inactive". This process is repeated until no active voxels can be removed without destroying the topological connectivity of the crack region. The center points of the remaining voxels after iterative erosion are extracted to form an initial skeleton point set. Smoothing filtering and redundant point removal are performed on the initial skeleton point set to ensure the continuity of the skeleton lines and the single-pixel width, generating a refined crack skeleton line point cloud. Optionally, smoothing filtering can be performed by resampling and smoothing the initial skeleton point set using the moving least squares method. Redundant point removal can check the connections between skeleton points. If the distance between two points is less than a set threshold and the direction of their connection deviates very little from the main direction of the local skeleton, then one of the points is removed to ensure that the skeleton line consists of only a single point on each cross-section.

[0023] See Figure 3 In one embodiment of the present invention, key point identification is performed on the crack skeleton line point cloud. Key points include endpoints, branch points, and intersection points of the skeleton line. Specifically, for a refined crack skeleton line point cloud extracted from a concrete wall surface, each point in the skeleton point cloud is traversed. The number of nearest neighbors for each point within a spherical neighborhood with a set search radius is checked. If the number of nearest neighbors is one, the point is identified as an endpoint; if the number of nearest neighbors is greater than two, the point is identified as a branch point or intersection point. To distinguish between branch points and intersection points, it is further analyzed whether the connection paths of the skeleton points in the neighborhood of that point branch into three or more independent extension directions in three-dimensional space. The three-dimensional Euclidean distance between all point pairs in the crack skeleton line point cloud is calculated, and the direct connection relationship between points is determined based on a preset distance threshold. In some embodiments, a connection distance threshold is set. For any two points in the skeleton point cloud, if the three-dimensional Euclidean distance between the two points is less than or equal to the connection distance threshold, and there is no third skeleton point between the two points such that its distance to both points is less than the connection distance threshold, then it is determined that there is a direct connection relationship between the two points. An undirected graph structure is constructed using the identified key points as nodes and the paths between key points consisting of continuous skeleton points as edges. In the specific implementation, a graph data structure is used to represent this structure. Each identified endpoint, branch point, and intersection point is assigned a unique node identifier. Each path connecting two key points and consisting of a series of continuous skeleton points in between is defined as an edge. Each edge records the identifiers of the two nodes it connects and the ordered sequence of all intermediate skeleton points it contains.

[0024] Redundant edge elimination and ring structure analysis are performed on the undirected graph structure to ensure that the undirected graph structure accurately reflects the true topological connectivity of the crack network, generating the final crack topological connectivity graph. Optionally, redundant edge elimination can check whether there are multiple edges directly connecting two nodes. If so, only one optimal edge is retained and the rest are deleted based on the average length of the paths represented by these edges or other geometric characteristics. Ring structure analysis needs to detect whether there are loops in the graph. If a loop is detected, the geometric characteristics of the edges constituting the loop are analyzed. Usually, in the actual physical form of the crack skeleton, key points may be missed at the intersection of loops. In this case, it is necessary to backtrack the original skeleton point cloud in the central region of the loop, re-identify and insert a branch point node, and decompose the loop into multiple edges originating from the new node. In the topological connectivity graph, each edge connecting two endpoints is traversed, and the sum of the line segment lengths between all continuous points on the skeleton path corresponding to the edge is calculated as the length of the crack branch. It can be understood that for each edge, its corresponding skeleton path consists of a series of ordered points. By calculating and summing the Euclidean distances between adjacent points, the length of the continuous spatial curve of the crack branch can be obtained.

[0025] For each skeleton point in the crack skeleton line point cloud, within its normal plane, a search is performed from the skeleton point along both sides of the normal direction until the boundary of the background point cloud subset is reached. The sum of the search distances on both sides is taken as the local crack width at the skeleton point. In specific implementation, for a point located on the crack skeleton, the local principal direction of the point and its neighboring skeleton points is calculated through principal component analysis. This principal direction is the tangent direction of the skeleton line at this point, and the plane perpendicular to this tangent direction is the normal plane. Within the normal plane, with the skeleton point as the center, rays are projected along two mutually perpendicular search directions within the normal plane, both perpendicular to the tangent. Collision detection is performed between the rays and the background point cloud subset to find the first background point to collide with. The distances from the central skeleton point to the two collision points are measured, and the sum of these two distances is defined as the local crack width at the skeleton point. The local crack widths at all skeleton points on a crack branch are statistically analyzed, and their average value is calculated as the average width of the crack branch. The maximum value among these is taken as the maximum width of the crack branch. In some embodiments, for an edge in the topological connection graph, the sequence of all skeleton points corresponding to the edge is obtained, the local width of the crack is calculated for each point in the sequence, the arithmetic mean of all local width values ​​of the crack is the average width of the crack branch, and the maximum value among all local width values ​​of the crack is the maximum width of the crack branch.

[0026] Calculate the principal direction vector of the crack branch skeleton line, and then calculate the angle between the principal direction vector and the reference plane of the structural surface extracted from the background point cloud subset, as the direction angle of the crack branch. Optionally, the principal direction vector can be obtained by performing principal component analysis on all skeleton points corresponding to the crack branch; the eigenvector corresponding to the largest eigenvalue is the principal direction vector. The direction angle of the crack branch is calculated by the angle between the principal direction vector and the normal vector of the reference plane of the structural surface, specifically defined as:

[0027] Where: symbol The calculated angle of the crack branch direction, represented by the symbol. The principal direction vector representing the crack branch skeleton line, symbol The symbol represents the unit normal vector of the reference plane of the structural surface extracted from the subset of the background point cloud. The dot product operation represents vectors, with the sign... The magnitude of the vector is represented. The length, average width, maximum width, and orientation angle of all crack branches are summarized to form a set of geometric parameters for the crack. This can be understood as traversing all edges in the crack topology graph, performing the above calculation process on each edge, and recording and compiling the length, average width, maximum width, and orientation angle of the corresponding crack branches for each edge to form a complete set of parameters describing the geometric morphology of the crack network.

[0028] In one embodiment of the present invention, a plane fitting process based on random sampling consensus is performed on a subset of background point clouds to extract the principal structural plane equation of the target structure. Specifically, for the subset of background point clouds segmented from the point cloud on the surface of the concrete bridge pier, the random sampling consensus algorithm is used to randomly select three points from the subset to define an initial plane model. Then, the distances from all points in the subset to this initial plane model are calculated, and the number of points whose distances are less than a set inlier threshold is counted. This number is taken as the number of support points for the current plane model. In a specific implementation, the above random sampling, model evaluation, and update process is iterated a preset number of times, and finally, the plane model with the most support points is selected as the best plane fitting result. The plane equation of the best plane fitting result can be expressed as: In the form of, These are the components of the plane normal vector. This is a constant term. It calculates the normal vector of the principal structural plane in three-dimensional space and the extent of the plane's boundary. The normal vector of the principal structural plane is directly obtained from the plane equation coefficients. We obtain, i.e., the normal vector. The planar boundary is determined by projecting all points within a subset of the background point cloud onto the principal structural plane and then calculating the minimum bounding rectangle of the projected point set in the planar two-dimensional space. In some embodiments, a local two-dimensional coordinate system is defined within the principal structural plane, with the projection center of the principal structural plane as the origin and the two mutually perpendicular principal directions determined by principal component analysis as the coordinate axes. The coordinates of all projected points in the new local two-dimensional coordinate system are recorded, and their minimum and maximum values ​​define the boundary of the plane.

[0029] Using the main structural plane as a reference, a local two-dimensional coordinate system is established, and points from the background point cloud subset are projected onto this local two-dimensional coordinate system to obtain the two-dimensional projection point set of the structural surface. In specific implementation, the process of establishing the local two-dimensional coordinate system is as follows: using the normal vector of the main structural plane... As a new local coordinate system An axis, arbitrarily chosen in the plane, is not parallel to... vector and The cross product yields a new vector that serves as the local coordinate system. Shaft, then use and The cross product of axial vectors yields the local coordinate system. An orthogonal coordinate system is thus constructed. Each 3D point in the background point cloud subset calculates its coordinates in the newly created local 2D coordinate system using a coordinate transformation formula. This transformation is essentially a projection process from the point to the main structural plane, resulting in a 2D projection point set of the structural surface. Based on this 2D projection point set, the flatness deviation and local undulation characteristics of the structural surface are calculated as supplementary information to the structural surface baseline. Optionally, the flatness deviation can be measured by calculating the statistical measure of the distance from each point in the 2D projection point set to the main structural plane, such as calculating the standard deviation of the distances from all points to the plane. Local undulation characteristics can be described by generating an elevation map through spatial interpolation of the 2D projection point set and then analyzing the texture features of the elevation map. In some embodiments, the roughness parameter of the 2D projection point set of the structural surface is calculated, and the formula for calculating the roughness parameter is defined as:

[0030] Where: symbol The root mean square roughness calculation result represents the two-dimensional projection point set of the structural surface, with the symbol... The symbol represents the total number of background point cloud sub-points. Representative background point cloud sub-collection No. The vertical distance from each point to the main structural plane. It can be understood that flatness deviation and local undulation characteristics are important supplementary information for evaluating the macroscopic morphology of the structural surface.

[0031] Each point in the crack skeleton point cloud is projected onto the main structural plane to obtain the two-dimensional projection trajectory of the crack on the structural surface. Specifically, the coordinates of each three-dimensional point in the crack skeleton point cloud are substituted into the equation of the main structural plane to calculate the coordinates of its perpendicular projection point. In the local two-dimensional coordinate system, the centroid coordinates of the crack's two-dimensional projection trajectory are calculated as the planar position of the crack on the structural surface. The centroid coordinates of the crack's two-dimensional projection trajectory are obtained by the arithmetic mean of the two-dimensional coordinates of all projection points on its boundary. The average distance from all points in the crack skeleton point cloud to the main structural plane is calculated as the average embedding depth of the crack relative to the structural surface. In practice, for each point in the crack skeleton point cloud, its perpendicular distance to the main structural plane is calculated, and then the arithmetic mean of all distance values ​​is taken; the result of the arithmetic mean is the average embedding depth. The angle between the crack's direction angle and the main direction of local surface undulations is calculated as the relative angle between the crack's direction and the surface texture. In some embodiments, the principal direction of local undulations on the structural surface is obtained through directional statistical analysis of the elevation map or point cloud normal variation of the two-dimensional projection point set, for example, by calculating the principal direction of the gradient direction histogram of the two-dimensional projection point set. The relative angle is the minimum angle between the crack orientation angle and this principal direction of local undulation. It can be understood that this angle reflects the spatial relationship between the crack extension direction and the inherent texture direction of the structural surface. The planar position of the crack, the average embedment depth, and the relative angle between the crack orientation and the surface texture together constitute the spatial position parameters of the crack.

[0032] In one embodiment of the present invention, a standardized crack measurement report data structure template is established. The data structure template includes fields for storing crack identifiers, geometric parameters, spatial location parameters, and measurement timestamps. In one example, The formatted data structure template includes the following top-level fields: a unique identifier for the measurement project, scanning equipment information, measurement time, and an array of "crack lists". The measurement equipment information field contains metadata such as the laser scanner's model, serial number, and scanning resolution. The measurement time field records the date the data acquisition was completed. Timestamp. Each element in the "Crack List" array represents an identified and analyzed crack. Each crack element contains multiple subfields for storing parameters calculated in the implementation example. The unique crack identifier field in the crack element is a globally unique identifier automatically generated by the system. The geometric parameter subfields in the crack element include length, average width, maximum width, and orientation angle. The spatial location parameter subfields in the crack element include planar location (two-dimensional coordinates), average embedding depth, and the relative angle between the crack orientation and the surface texture. The crack element also contains a field for associating the path to the graphical file or embedded data storing the crack topology connection diagram.

[0033] The calculated set of crack geometry parameters and crack spatial location parameters are populated according to the field format of the data structure template. In specific implementation, for an analysis result containing three branch cracks, the system generates three crack elements, which are then populated into the "Crack List" array. The values ​​of length, average width, maximum width, and direction angle in the crack geometry parameter set are assigned to the corresponding "Length," "Average Width," "Maximum Width," and "Direction Angle" fields of each crack element, respectively. The values ​​of planar position coordinates, average embedding depth, and relative angle in the crack spatial location parameters are assigned to the corresponding "Planar Position," "Average Embedding Depth," and "Relative Angle" fields of each crack element, respectively. This structured filling ensures that the quantitative information of each crack is recorded accurately and independently. Before filling, the data of the crack geometry parameter set and crack spatial location parameters can be formatted according to precision requirements, for example, all floating-point numbers are uniformly retained to three decimal places.

[0034] The crack topology connectivity map is encoded in vector graphics format and associated with the corresponding crack identifiers. Optionally, the crack topology connectivity map can be encoded as a scalable vector graphic. Format or drawing exchange file Format. During the encoding process, the structural surface reference plane extracted from the background point cloud subset is used as the drawing reference plane. The crack skeleton line point cloud and its key points (endpoints, branch points, intersections) are projected onto this reference plane. Different crack branches are represented by line segments of different colors and line types, and key points are marked with specific symbols. The encoded vector graphics file can be stored independently and associated with a "graphic reference" field in the crack element of the "crack list" array in the data structure template through the crack identifier. This field stores the path of the graphics file or contains the graphics data. Encoded strings. This association method allows data in measurement reports to be correlated and retrieved with their graphical representations. In some embodiments, to further compress data volume, vector graphics data can be encoded using a lossless compression algorithm before storage. The association reference relationship is implemented through a query function, the formula of which is defined as:

[0035] Where: symbol Represents the generated query key value, symbol The unique crack identifier, symbol representing a crack. This represents a hash function. This query key-value pair... It can be used to quickly locate and identify cracks in graphic file indexes or embedded data blocks. The corresponding crack topology connection diagram data.

[0036] All the filled and coded data are integrated to generate a structured data file, namely the comprehensive measurement report. In practice, the integration process packages the filled data structure instance and associated graphic files according to a predefined encapsulation format. Optionally, a compressed file package conforming to a specific standard can be generated, containing a master data file and one or more graphic files. The master data file is linked to the graphic files within the package through reference relationships. This structured comprehensive measurement report file can be directly read, parsed, and processed by subsequent computer-aided assessment systems, structural health monitoring platforms, or databases. The table shows an example of the filled content for a crack element in the data structure template.

[0037] See Figure 4 Using a box plot visualization paradigm, the laser measurement system's identification data is accurately presented. The statistical distribution characteristics of the local width of three target cracks are derived from the calculation results of the local width in the normal plane of the crack skeleton line point cloud. The chart uses the globally unique identifier of the crack as the horizontal axis for classification, and the vertical axis represents the local width of the crack (unit: m), comprehensively covering the full-dimensional statistical information of the minimum, lower quartile (Q1), median (dashed line), upper quartile (Q3), maximum, and outliers. The upper and lower edges of the box correspond to Q3 and Q1 respectively, and the height of the box represents the interquartile range (IQR), reflecting the core dispersion of the crack width. The red dashed line inside the box represents the median, indicating the concentration trend of the local crack width. The whisker extends to the extreme values ​​within 1.5 times the IQR, and the scattered points outside the whisker represent abnormal width values ​​exceeding this range, corresponding to extreme local expansion or measurement feature points of the crack. From a quantitative perspective… and The width distribution of both cracks exhibits low values ​​and small dispersion, with medians of approximately 0.02 m and interquartile ranges less than 0.01 m, and no significant outliers. This indicates that the two cracks have a uniform local width distribution and strong geometric stability. The width distribution exhibits high numerical values ​​and large dispersion, with a median of approximately 0.10 m and an interquartile range exceeding 0.04 m. It also shows outliers with an upper limit of approximately 0.17 m and a lower limit of approximately 0.035 m, with the upper edge of the crack reaching 0.14 m and the lower edge reaching 0.06 m. This accurately quantifies the local width heterogeneity of the crack and is highly consistent with the calculation results of "local width of skeleton points" in the laser measurement method. This provides core visualization support for subsequent statistical analysis of crack geometric parameters (average width, maximum width) and structural health risk assessment.

[0038] In one embodiment of the invention, design drawings or a rough 3D model of the target structure are acquired to identify crack-prone areas requiring focused monitoring. In a specific implementation, for crack measurement of a large concrete bridge, the design drawings of the target structure are either 2D drawings in computer-aided design format or 3D model files in building information model format. By analyzing the structural component connections, stress concentration areas (such as near supports and mid-span sections) in the design drawings, and the locations of common defects recorded in historical inspection reports, the tops of the bridge piers, the junctions of the main beam web and flanges, and the areas near expansion joints are identified as crack-prone areas requiring focused monitoring. The identification process can be based on structural mechanics knowledge, material fatigue characteristics, and previous maintenance records.

[0039] Based on the spatial distribution of easily cracked areas, the scanning path and station layout of the laser measuring instrument are planned to ensure comprehensive coverage of the target area. In practice, a terrestrial 3D laser scanner is used. The planning process takes place in 3D space, firstly by spatially locating and marking the bounding boxes of the identified easily cracked areas within the overall coordinate system of the target structure. When planning the scanning path and station layout, the laser scanner's field of view, effective range, and overlap requirements between adjacent stations must be considered. For the easily cracked area of ​​the bridge pier top, the station layout plan involves setting up multiple scanning stations on the ground around the pier and on the adjacent bridge deck to ensure comprehensive scanning of the pier's top and sides from different angles. The combined scanning range of all stations completely encompasses the bounding box of the pier top. The scanning path is planned to control the scanner to automatically move according to the order in the station layout plan and complete the scanning at each station. The overlap rate of the scanning areas of adjacent stations is no less than 30% to ensure seamless stitching of the point cloud data. This planning allows for the systematic acquisition of high-quality 3D point cloud data covering all easily cracked areas.

[0040] Based on the surface material of the target structure and the expected crack size, the scanning resolution and sampling frequency parameters of the laser measuring instrument are set. The surface material of the target structure affects the reflectivity of the laser and the effective measurement distance, while the expected crack size determines the required spatial resolution. For a concrete structure surface with an expected crack width greater than 0.1 mm, the scanning resolution parameter of the laser measuring instrument needs to be set to a value sufficient to resolve this width. The scanning resolution parameters of the laser measuring instrument include angular resolution (horizontal and vertical directions) and ranging accuracy. In practical implementation, the angular resolution setting needs to ensure that at the maximum effective measurement range, the minimum interval between adjacent laser points is less than half of the expected minimum crack width, so as to ensure that the crack can be represented by at least two points in the point cloud data. The sampling frequency parameter of the laser measuring instrument determines the density and speed of point cloud data acquisition, and a balance needs to be struck between measurement accuracy and operational efficiency. For general surveys of large structures, a lower sampling frequency can be used to quickly obtain an overview, while for key crack-prone areas, the highest sampling frequency is used to obtain a high-density point cloud. During the planning process, the calculation formulas for the scanning resolution and sampling frequency parameters are defined as follows:

[0041] Where: symbol Represents the laser measuring instrument at distance Theoretical point spacing, sign Represents the measurement distance from the laser measuring instrument to the surface of the target structure, symbol This represents the angular resolution (in radians) of the laser measuring instrument. When setting the scanning resolution and sampling frequency parameters, it is necessary to ensure that the measurement is performed at the maximum distance. Theoretical point spacing calculated at [location] It is less than half the target expected crack width.

[0042] Based on the planned scanning path and site layout, the laser measuring instrument automatically performs area scanning measurement tasks. In some embodiments, the laser measuring instrument integrates an automatic control system or can receive external control commands. The planned scanning path and site layout are converted into a series of executable commands, including the 3D coordinates of each site, the scanner's attitude angle (pitch, yaw), and the scanning resolution and sampling frequency parameters for that site. Optionally, the control commands can be sent to the laser measuring instrument's control unit via a wireless network or wired connection. The laser measuring instrument automatically moves to the designated site coordinates according to the commands, adjusts its attitude angle, and scans the area covered by that site according to the preset scanning resolution and sampling frequency parameters, acquiring the local 3D point cloud data of that site. After the scanning tasks of all sites are completed, the system automatically, or after operator confirmation, initiates the point cloud data stitching process, registering the local point cloud data of each site to a unified global coordinate system through target or feature matching, ultimately generating a high-density 3D point cloud dataset for subsequent processing.

[0043] See Figure 5 In the risk assessment and crack statistical analysis of bridge crack-prone areas, a dual-axis visualization system was used to quantitatively characterize the structural defects. The left vertical axis represents the risk level (0-100), presenting the structural safety risk level of each area in the form of a bar chart; the right vertical axis represents the number of historical cracks, reflecting the frequency of defects in each area in the form of a line graph. From the data distribution, the top area of ​​the pier has the highest risk level (approximately 95), with 28 historical cracks, indicating significant potential risks related to structural stress concentration and material fatigue in this area. The risk level at the junction of the main beam web and flange is the second highest (approximately 90), but the number of historical cracks reaches a peak of 35, reflecting the fatigue cracking characteristics of this part under repeated loading. The area near the expansion joint has a risk level of approximately 85, with 18 historical cracks, reflecting the influence of temperature stress and structural deformation on this area. The bearing area has a risk level of approximately 80, with 22 historical cracks, related to bearing reaction concentration and constraint deformation. The mid-span section has a risk level of approximately 75, with 15 historical cracks, mainly affected by positive bending moment. The dual-axis visualization model enables collaborative analysis of risk levels and historical crack data. Through the coupling relationship between bar charts and line charts, the key control areas of "high risk - high cracks" can be intuitively identified, providing data support for the site deployment and sampling frequency setting of subsequent laser measurement schemes.

[0044] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A smart method for measuring structural cracks based on laser measurement, characterized in that, The method includes: A laser measurement instrument is used to perform regional scanning measurement on the surface of the target structure to obtain a high-density three-dimensional point cloud dataset containing crack information. Perform surface structure feature segmentation processing on the high-density three-dimensional point cloud dataset to separate the background point cloud subset corresponding to the complete structure surface and the candidate point cloud subset corresponding to the suspected crack region; Geometric morphological thinning and skeleton extraction are performed on the candidate point cloud subset to generate crack skeleton line point cloud, and topological connectivity analysis is performed on the crack skeleton line point cloud to construct the crack topological connectivity graph. Based on the crack skeleton line point cloud and the topological connection map, the set of geometric morphological parameters of the crack is calculated, including crack length, average width, maximum width and orientation angle. The set of geometric morphology parameters is fused with the structural surface reference information extracted from the background point cloud subset to calculate the spatial position parameters of the crack relative to the structural surface, generating a comprehensive measurement report that includes the crack's geometry and spatial position.

2. The intelligent structural crack measurement method based on laser measurement according to claim 1, characterized in that, Surface structure feature segmentation processing is performed on the high-density 3D point cloud dataset to separate the background point cloud subset corresponding to the intact surface structure and the candidate point cloud subset corresponding to the suspected crack region, including: Calculate the local curvature features and normal vector of each point in the high-density 3D point cloud dataset, and identify potential discontinuities based on the degree of abrupt change in the local curvature features and normal vector; For each potential discontinuity point, a local neighborhood point set is constructed, and the eigenvalues ​​of the three-dimensional spatial distribution covariance matrix of the local neighborhood point set are calculated. Based on the ratio relationship of the aforementioned feature values, linear distribution features and planar distribution features are distinguished, and local neighborhood point sets exhibiting obvious linear distribution features are initially marked as crack candidate points. A region growing operation is performed on the initially marked crack candidate points to merge spatially adjacent crack candidate points with the same normal direction, forming a continuous set of suspected crack region point clouds, i.e., the candidate point cloud subset. The candidate point cloud subset is removed from the original high-density 3D point cloud dataset, and the remaining point clouds constitute the background point cloud subset.

3. The intelligent structural crack measurement method based on laser measurement according to claim 2, characterized in that, Geometric morphological thinning and skeleton extraction are performed on the candidate point cloud subset to generate a crack skeleton line point cloud, including: The candidate point cloud subset is subjected to voxel rasterization downsampling processing to generate a regularized three-dimensional voxel mesh model; In the three-dimensional voxel mesh model, an iterative erosion operation is performed on the voxels marked as crack regions until the remaining voxels form a linear structure with a single voxel width. The remaining voxel center points after iterative erosion are extracted to form the initial skeleton point set; The initial skeleton point set is subjected to smoothing filtering and redundant point removal to ensure the continuity of the skeleton lines and the width of a single pixel, thereby generating a refined crack skeleton line point cloud.

4. The intelligent structural crack measurement method based on laser measurement according to claim 3, characterized in that, Perform topological connectivity analysis on the crack skeleton line point cloud to construct a topological connectivity graph of the crack, including: Key points are identified in the point cloud of the crack skeleton line, and the key points include the endpoints, branch points and intersections of the skeleton line. Calculate the three-dimensional Euclidean distance between all point pairs in the crack skeleton line point cloud, and determine the direct connection relationship between points based on a preset distance threshold; Using the identified key points as nodes and the paths formed by continuous skeleton points between key points as edges, an undirected graph structure is constructed. Redundant edge elimination and ring structure parsing are performed on the undirected graph structure to ensure that the undirected graph structure can accurately reflect the real topological connection relationship of the crack network, and the final crack topological connection graph is generated.

5. The intelligent structural crack measurement method based on laser measurement according to claim 4, characterized in that, Based on the crack skeleton line point cloud and the topological connectivity graph, the set of geometric morphological parameters of the crack is calculated, including: In the topological connection graph, each edge connecting two endpoints is traversed, and the sum of the line segment lengths between all consecutive points on the skeleton path corresponding to the edge is calculated as the length of the crack branch. For each skeleton point in the crack skeleton line point cloud, in its normal plane, start from the skeleton point and search along both sides along the normal direction until the boundary of the background point cloud subset is reached, and the sum of the search distances on both sides is taken as the local width of the crack at the skeleton point. The local width of the crack at all skeleton points on a crack branch is statistically analyzed, and the average value is calculated as the average width of the crack branch. The maximum value among them is taken as the maximum width of the crack branch. Calculate the principal direction vector of the crack branch skeleton line, and calculate the angle between the principal direction vector and the structural surface reference plane extracted from the background point cloud subset, as the direction angle of the crack branch; The length, average width, maximum width, and orientation angle of all crack branches are summarized to form a set of geometric parameters for the crack.

6. The intelligent structural crack measurement method based on laser measurement according to claim 5, characterized in that, The structural surface reference information extracted from the background point cloud subset includes: A plane fitting process based on random sampling consistency is performed on the background point cloud subset to extract the principal structure plane equation of the target structure; Calculate the normal vector of the main structural plane in three-dimensional space and the range of the plane boundary; Using the main structural plane as a reference, a local two-dimensional coordinate system is established, and the points in the background point cloud subset are projected onto the local two-dimensional coordinate system to obtain a two-dimensional projection point set of the structural surface; Based on the two-dimensional projection point set, the flatness deviation and local undulation characteristics of the structural surface are calculated as a supplement to the reference information of the structural surface.

7. The intelligent structural crack measurement method based on laser measurement according to claim 6, characterized in that, The set of geometric morphological parameters is fused with the structural surface reference information extracted from the background point cloud subset to calculate the spatial position parameters of the crack relative to the structural surface, including: Project each point in the crack skeleton line point cloud onto the main structural plane to obtain the two-dimensional projection trajectory of the crack on the structural surface. In the local two-dimensional coordinate system, the centroid coordinates of the two-dimensional projection trajectory of the crack are calculated as the planar position of the crack on the surface of the structure. Calculate the average distance from all points in the crack skeleton line point cloud to the main structural plane, and use it as the average embedment depth of the crack relative to the structural surface; Calculate the angle between the crack direction angle and the main direction of local undulations on the structural surface, and use it as the relative angle between the crack direction and the surface texture. The spatial position parameters of the crack are constituted by the planar position of the crack, the average embedment depth, and the relative angle between the crack direction and the surface texture.

8. The intelligent structural crack measurement method based on laser measurement according to claim 7, characterized in that, The generation of a comprehensive measurement report, including the crack geometry and spatial location, includes: Establish a standardized crack measurement report data structure template, which includes fields for storing crack identifiers, geometric parameters, spatial location parameters, and measurement timestamps; The calculated set of geometric morphological parameters of the crack and the spatial location parameters of the crack are filled in according to the field format of the data structure template; The crack topology connection diagram is encoded in vector graphics format and associated with the corresponding crack identifier; All the filled and coded data are integrated to generate a structured data file, namely the comprehensive measurement report.

9. The intelligent structural crack measurement method based on laser measurement according to claim 8, characterized in that, Before performing area scanning measurement of the target structure surface using a laser measuring instrument, a measurement planning step is also included: Obtain design drawings or a rough 3D model of the target structure and identify the crack-prone areas that require key monitoring; Based on the spatial distribution of the easily cracked areas, the scanning path and station layout of the laser measuring instrument are planned to ensure complete coverage of the target area. Based on the surface material of the target structure and the expected crack size, set the scanning resolution and sampling frequency parameters of the laser measuring instrument; Based on the planned scanning path and site layout, the laser measuring instrument is controlled to automatically perform the area scanning measurement task.

10. A laser-based intelligent structural crack measurement system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent structural crack measurement method based on laser measurement as described in any one of claims 1 to 9.

Images