Three-dimensional laser scanning beam density precision regulation and control method based on galvanometer dynamic regulation and control

Abstract

The application discloses a three-dimensional laser scanning beam density precision regulation and control method based on a galvanometer dynamic regulation and control, and belongs to the technical field of three-dimensional laser scanning. The method comprises the following steps: performing initial three-dimensional laser scanning on a target region to obtain initial point cloud data; distributing grid coordinates to the point cloud; dynamically setting laser beam density and beam spacing according to parameter identification region types; analyzing the region according to the set parameters to generate a spiral scanning path; converting the path into a galvanometer control signal to control the laser beam to scan along the path; and precisely regulating laser parameters according to the scanning result. The method solves the technical problem that traditional three-dimensional laser scanning cannot adjust the spatial distribution density of the light beam in real time and cannot meet the differentiated precision structure surface complete scanning of the scene. The method realizes three-dimensional laser scanning beam density precision regulation and control, improves scanning precision and efficiency, reduces redundant data, and meets the scanning requirements of roadway surrounding rock, geological structure and other scenes.

Description

Technical Field

[0001] This invention belongs to the field of three-dimensional laser scanning technology, specifically relating to a method for precise control of beam density in three-dimensional laser scanning based on dynamic control of galvanometers. Background Technology

[0002] 3D laser scanning technology can quickly acquire the 3D coordinates of the surface of an object under test. The resulting point cloud data is highly accurate, dense, and contains rich semantic information such as reflection intensity. This technology constructs surface point cloud data of the object under test through the spatial projection and reflection signal acquisition of a laser beam, thereby achieving accurate reconstruction of a 3D model. In special application scenarios such as tunnel surrounding rock monitoring and geological structure analysis, the objects under test often have highly complex surface features. For example, the edges of rock mass structures may have millimeter-level cracks, while intact rock mass areas exhibit relatively uniform geometry. This extreme difference in surface geometry places almost contradictory demands on the scanning system: achieving sub-millimeter-level high-precision resolution in critical areas while avoiding data redundancy in non-critical areas.

[0003] The current mainstream 3D laser scanning system faces a dilemma: fixed-density scanning modes are difficult to adapt to these differentiated needs. While global high-density scanning can acquire fine features of structural edges, it results in a large amount of redundant data in the intact rock mass area, significantly increasing data processing load and storage costs. Conversely, while global low-density scanning can improve overall efficiency, it is difficult to obtain accurate structural information at rock mass edges and in fractured areas. Summary of the Invention

[0004] This invention provides a method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer adjustment. This method addresses the technical problem that traditional three-dimensional laser scanning cannot adjust the spatial distribution density of the beam in real time, making it difficult to meet the requirements for complete scanning of differentiated structural surfaces in various scenarios. By acquiring point cloud data and dividing the target area into grid cells through initial three-dimensional laser scanning, calculating local feature parameters of the cells to identify the region type and dynamically setting beam parameters, generating a spiral scanning path and converting it into a galvanometer control signal, and combining the scanning results to precisely control the beam parameters, the method achieves precise control of beam density in three-dimensional laser scanning, improving scanning accuracy and efficiency while reducing redundant data, thus meeting the scanning needs of scenarios such as roadway surrounding rock and geological structures.

[0005] In view of the above problems, the present invention adopts the following technical solution:

[0006] This invention discloses a method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer control. The method includes: performing an initial three-dimensional laser scan on a target area to acquire initial point cloud data; dividing the initial point cloud data into multiple spatial grid cells and calculating local feature parameters of the multiple spatial grid cells; identifying the region type of the multiple spatial grid cells based on the local feature parameters, and dynamically setting the laser beam density and beam spacing based on the identification results; analyzing the regions in the identification results according to the laser beam density and the beam spacing to generate a spiral scanning path; converting the spiral scanning path into a galvanometer control signal, controlling the laser beam to scan along the spiral scanning path through the galvanometer control signal, and precisely controlling the laser beam density and the beam spacing based on the scanning results.

[0007] Preferably, the present invention also discloses a method for dividing the initial point cloud data into multiple spatial grid units, comprising: identifying the extreme values ​​of the initial point cloud data according to the three-dimensional coordinate axes to determine the spatial range of the point cloud; setting the grid unit size, dividing the three-dimensional coordinate axes according to the grid unit size to determine the number of grids; using the point cloud spatial range as a constraint, dividing the initial point cloud data according to the number of grids to generate the multiple spatial grid units.

[0008] Preferably, the present invention also discloses a method for calculating the local feature parameters of the plurality of spatial grid cells, comprising: traversing the plurality of spatial grid cells to calculate the center point and determine the coordinates of the plurality of center points; extracting the coordinates of the plurality of points of the plurality of spatial grid cells and calculating the mean according to the three-dimensional coordinate axes to determine the centroid of the plurality of spatial grid cells; calculating the deviation between the plurality of point coordinates and the centroid to obtain the deviation vector and construct a covariance matrix; using the coordinates of the plurality of center points as the reference position to perform eigenvalue decomposition on the covariance matrix to obtain the plurality of eigenvalues ​​and normalizing them to obtain normalized eigenvalues; performing parallel calculation based on the normalized eigenvalues ​​and the plurality of spatial grid cells to obtain the information entropy and the surface curvature of the plurality of spatial grid cells; and integrating the information entropy and the surface curvature of the plurality of spatial grid cells to construct the local feature parameters.

[0009] Preferably, the present invention also discloses a method for identifying the region type of multiple spatial grid cells based on the local feature parameters, and dynamically setting the laser beam density and beam spacing based on the identification results. The method includes: identifying the region type of multiple spatial grid cells based on the local feature parameters, generating an identification result, wherein the identification result includes multiple region types; traversing multiple spatial grid cells for matching based on the multiple region types, generating a matching result; and dynamically setting the laser beam density and beam spacing of the multiple spatial grid cells according to the matching result.

[0010] Preferably, the present invention also discloses a method for identifying the region types of multiple spatial grid cells based on the local feature parameters and generating identification results, wherein the identification results include multiple region types. The method includes: setting a curvature threshold based on the surface curvature of the multiple spatial grid cells; performing extreme value analysis based on the information entropy of the multiple spatial grid cells to set a first information entropy threshold and a second information entropy threshold; comparing the surface curvature of the multiple spatial grid cells with the curvature threshold and the information entropy of the multiple spatial grid cells with the first information entropy threshold to determine whether the surface curvature of any grid cell in the multiple spatial grid cells is greater than the curvature threshold or the surface curvature of any grid cell in the multiple spatial grid cells is greater than the curvature threshold; and determining whether the surface curvature of any grid cell in the multiple spatial grid cells is greater than the curvature threshold or whether the surface curvature of any grid cell in the multiple spatial grid cells is greater than the curvature threshold. If the information entropy of a given element is greater than the first information entropy threshold, then the first grid cell is extracted as the edge region of the structural surface. A quality factor evaluation is performed based on multiple spatial grid cells, setting a first quality factor range and a second quality factor threshold. If the quality factor of any grid cell in the multiple spatial grid cells falls within the first quality factor range, then the second grid cell is extracted as the fractured rock mass region. If the information entropy of any grid cell in the multiple spatial grid cells is less than the second information entropy threshold, then the third grid cell is extracted as the intact rock mass region. The edge region of the structural surface, the fractured rock mass region, and the intact rock mass region are categorized to determine the types of the multiple regions.

[0011] Preferably, the present invention also discloses a method for analyzing the region in the identification result according to the laser beam density and the beam spacing to generate a spiral scanning path, comprising: identifying the region according to the multiple region types to determine the region to be scanned; mapping the region to be scanned to multiple spatial grid cells to determine multiple target grid cells; extracting the center point of the multiple target grid cells as the starting center point; performing spiral scanning analysis based on the density of the multiple target points of the multiple target grid cells to generate spiral path parameters; and performing a dense scan of the region to be scanned from the starting center point in combination with the spiral path parameters based on the laser beam density and the beam spacing to generate the spiral scanning path.

[0012] Preferably, the present invention also discloses a method for converting the helical scanning path into a galvanometer control signal, and controlling the laser beam to scan along the helical scanning path using the galvanometer control signal. The method includes: traversing all path points within the scanning path and marking the points to determine multiple path points, wherein the scanning path includes a helical scanning path generated for the edge of the identified structural surface; extracting the three-dimensional Cartesian coordinates of the multiple path points and converting them to generate a galvanometer deflection angle; constructing a galvanometer control signal based on the galvanometer deflection angle; controlling the laser beam to traverse the scanning points of the scanning path using the galvanometer control signal to obtain a scanning result, wherein the scanning result includes a control signal sequence; and precisely controlling the laser beam density and the beam spacing based on the control signal sequence.

[0013] Preferably, the present invention also discloses a method for precisely controlling the laser beam density and the beam spacing based on the control signal sequence, comprising: setting a desired point cloud density threshold; extracting point cloud scanning data based on the control signal sequence; comparing the point cloud scanning data with the desired point cloud density threshold; if there is a deviation between the point cloud scanning data and the desired point cloud density threshold, calculating the deviation data; adaptively adjusting the laser beam density and the beam spacing according to the deviation data to generate an adjustment result; and updating the control signal sequence according to the adjustment result to generate an optimized control signal sequence.

[0014] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0015] An initial 3D laser scan of the target area is performed to acquire initial point cloud data. This initial point cloud data is then divided into multiple spatial grid cells, and local feature parameters of these cells are calculated. Based on these local feature parameters, region type identification is performed on the multiple spatial grid cells, and the laser beam density and beam spacing are dynamically set based on the identification results. The regions in the identification results are analyzed according to the laser beam density and beam spacing to generate a spiral scanning path. This spiral scanning path is converted into a galvanometer control signal, which controls the laser beam to scan along the spiral scanning path. The laser beam density and beam spacing are precisely adjusted based on the scanning results. This achieves precise control of the 3D laser scanning beam density, improving scanning accuracy and efficiency while reducing redundant data, thus meeting the scanning needs of scenarios such as tunnel surrounding rock and geological structures. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer control according to the present invention.

[0017] Figure 2This is a flowchart illustrating the dynamic density decision-making process of the three-dimensional laser scanning beam density precision control method based on galvanometer dynamic control of the present invention.

[0018] Figure 3 This is a schematic diagram of the on-site identification area for the three-dimensional laser scanning beam density precision control method based on galvanometer dynamic control of the present invention;

[0019] Figure 4 This is a flowchart of the spiral scanning path generation process of the three-dimensional laser scanning beam density precision control method based on galvanometer dynamic control of the present invention.

[0020] Figure 5 The flowchart of the galvanometer control system is shown in the present invention, which describes a method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer adjustment. Detailed Implementation

[0021] This invention provides a method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer adjustment. This solves the technical problem that traditional three-dimensional laser scanning cannot adjust the spatial distribution density of the beam in real time, making it difficult to meet the requirements of complete scanning of differentiated structural surfaces in different scenarios. It achieves precise control of beam density in three-dimensional laser scanning, improves scanning accuracy and efficiency, and reduces redundant data, thus meeting the scanning needs of scenarios such as roadway surrounding rock and geological structures.

[0022] like Figure 1 As shown, this invention provides a method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer adjustment, the method comprising:

[0023] S100: Perform initial 3D laser scanning on the target area to obtain initial point cloud data.

[0024] Specifically, this involves using a 3D laser scanning device to conduct an initial scan of the target area, including the surrounding rock and geological structure of the tunnel. This process collects and generates initial point cloud data that provides a preliminary 3D representation of the area's morphology, offering fundamental data support for subsequent control and management. During operation, the device's mounting position and angle are determined based on the actual area and environmental conditions, such as the presence of obstructions, ensuring complete coverage of the target area and avoiding blind spots. Simultaneously, reasonable scanning parameters, such as laser emission frequency and sampling interval, are preset according to the needs of subsequent mesh generation and feature analysis. This ensures the initial data accurately represents the overall contour of the area, including the approximate locations of protrusions, depressions, and fissures on the rock surface, avoiding redundancy due to excessively high parameters. The scanning device is then activated, continuously emitting laser beams towards the target area. Upon contact with the target surface, the laser beams are reflected, and the receiving module captures the reflected signals in real time. Based on the time difference, phase difference, or triangulation principles of laser propagation, the device calculates the coordinates (X, Y, Z) of each laser spot in 3D space. A large number of coordinate points are stored and integrated according to the scanning sequence, ultimately forming the initial point cloud data.

[0025] S200: Divide the initial point cloud data into multiple spatial grid cells and calculate the local feature parameters of the multiple spatial grid cells.

[0026] Specifically, the initial point cloud data is broken down into multiple spatial grid cells according to rules, and then the local feature parameters of each grid cell are calculated one by one. These local feature parameters include information entropy and curvature. Information entropy reflects the degree of disorder in the point cloud distribution within the cell; a higher value indicates a more irregular point cloud distribution and a more complex structure in the corresponding region. Curvature reflects the degree of unevenness of the target surface corresponding to the cell; a higher value indicates a steeper surface or the presence of obvious protrusions or depressions.

[0027] When dividing the point cloud into grid cells, the spatial distribution range of the point cloud is first determined based on the extreme values ​​of the initial point cloud data on the three-dimensional coordinate axes, thus defining the boundaries of the data on the coordinate axes. Then, based on the structural characteristics of the target area and the accuracy requirements of subsequent analysis, an appropriate grid cell size is set, and the three-dimensional coordinate axes are divided according to this size to determine the number of grid cells. Finally, with the spatial range of the point cloud as a constraint, the initial point cloud data is evenly divided according to the number of grid cells to generate multiple independent spatial grid cells, ensuring that each cell contains sufficient and non-redundant point cloud data.

[0028] When calculating local feature parameters, each grid cell is traversed first, the coordinates of the cell center point are calculated, and the coordinates of all points within the cell are extracted. The centroid of the cell is then calculated using the mean of the coordinates. Next, the deviation between the coordinates of each point within the cell and the centroid is calculated, and a covariance matrix is ​​constructed. The matrix is ​​then decomposed into eigenvalues ​​and normalized using the coordinates of the center point as a reference. Finally, the information entropy and curvature are calculated by combining the normalized eigenvalues.

[0029] S300: Based on the local feature parameters, perform region type identification on multiple spatial grid cells, and dynamically set the laser beam density and beam spacing based on the identification results.

[0030] Specifically, this involves determining the region type of each spatial grid cell based on its local characteristic parameters, and then setting the laser beam density and beam spacing accordingly for different region types. The identification result refers to the classification of each spatial grid cell using preset judgment criteria, including the region type. The preset judgment criteria are the criteria for region type identification based on the local characteristic parameters and quality factors of the spatial grid cells. These criteria include: a curvature threshold based on surface curvature; a first and second information entropy threshold based on information entropy extreme value analysis; and a first and second quality factor interval based on quality factor evaluation, used to determine the region type of the grid cell.

[0031] S400: Analyze the region in the recognition result according to the laser beam density and the beam spacing to generate a spiral scanning path.

[0032] Specifically, based on the identification results, the corresponding scanning areas for each type are determined. These areas are then precisely mapped onto multiple pre-defined spatial grid cells, identifying the target grid cells to be scanned. Next, the center point of each target grid cell is extracted as the starting center point for the spiral scan, providing a starting reference for the scanning path. Subsequently, the target point density is analyzed based on the actual distribution of the point cloud within the target grid cell. This analysis is used as the basis for spiral scan analysis, determining relevant parameters of the spiral path, such as the number of spiral turns and the step size per turn. Finally, based on the laser beam density and beam spacing previously set for different regions, starting from the starting center point and combining the determined spiral path parameters, an adaptive and denser scanning plan is implemented for each scanning region. This ensures that the distribution of scanning points meets the beam density and spacing requirements, ultimately generating a spiral scan path with complete coverage and appropriate accuracy.

[0033] S500: The spiral scanning path is converted into a galvanometer control signal, and the laser beam is controlled to scan along the spiral scanning path by the galvanometer control signal. The laser beam density and the beam spacing are precisely adjusted according to the scanning results.

[0034] Specifically, the generated spiral scanning path is converted into a control signal that the galvanometer can recognize. This signal is then used to control the laser beam to scan along the spiral scanning path. Finally, the laser beam density and beam spacing are further precisely adjusted based on the scanning results.

[0035] The present invention also discloses a method for dividing the initial point cloud data into multiple spatial grid units, comprising: identifying the extreme values ​​of the initial point cloud data according to the three-dimensional coordinate axes to determine the spatial range of the point cloud; setting the grid unit size, dividing the three-dimensional coordinate axes according to the grid unit size to determine the number of grids; using the spatial range of the point cloud as a constraint, dividing the initial point cloud data according to the number of grids to generate the multiple spatial grid units.

[0036] Specifically, this involves determining the spatial extent of the point cloud, setting the grid cell size, determining the number of grids, and then dividing the initial point cloud data into multiple spatial grid cells, constrained by the spatial extent of the point cloud. Extreme value identification refers to extracting the maximum and minimum values ​​along the X, Y, and Z axes from all the three-dimensional coordinate points contained in the initial point cloud data.

[0037] First, calculate the point cloud bounding box using the formula. , This involves determining the minimum and maximum vertices of the point cloud along the X, Y, and Z coordinate axes, thereby defining the spatial extent of the point cloud. A bounding box is the smallest rectangular box that can completely enclose all coordinate points of the initial point cloud data in three-dimensional space. Its minimum and maximum vertices correspond to the extreme combinations of the point cloud data along the X, Y, and Z coordinate axes, respectively. It refers to the smallest vertex of the bounding box of the point cloud along the three coordinate axes. It refers to the largest vertex of the bounding box of the point cloud in the three coordinate axis directions; Represents the X-axis coordinates of all point clouds Take the minimum value, where k For the index of point clouds, It is the coordinate value of the k-th point on the X-axis, and the result is the minimum boundary in the X-axis direction; This indicates taking the minimum value of the Y-axis coordinate for all point clouds. It is the coordinate value of the k-th point on the Y-axis, and the result is the minimum boundary in the Y-axis direction; Represents the Z-axis coordinates of all point clouds Take the minimum value. It is the coordinate value of the k-th point on the Z-axis, and the result is the minimum boundary in the Z-axis direction; This indicates that the X-axis coordinates of all point clouds are taken as follows. The maximum value is the maximum boundary in the X-axis direction. This indicates that the Y-axis coordinates of all point clouds are taken as... The maximum value is the maximum boundary in the Y-axis direction. This indicates that the Z-axis coordinates of all point clouds are taken as... The maximum value is the maximum boundary in the Z-axis direction.

[0038] Next, the grid dimensions are determined, and the grid cell size is set according to the structural complexity of the target area and the required analysis accuracy. Through formula

[0039] ,

[0040] Calculate the number of grid cells along each coordinate axis. Since the grid covers a closed interval, the number of grid cells equals the interval length divided by the cell size, rounded down, and then incremented by 1 to include the boundary. i Corresponding to the X, Y, and Z axes respectively. Indicates the number of grid cells. and These represent the point cloud at the th... i Maximum and minimum values ​​on each coordinate axis (X, Y, Z); It is the size of the grid cell; This indicates rounding down to the nearest integer.

[0041] For example, taking a roadway surrounding rock scanning scenario as an example, the scanning area is a 5m×5m roadway rock wall. The specific implementation steps are as follows: First, the system is initialized by installing the 3D laser scanner on the side wall of the roadway, 1.5m away from the scanning surface; the parameters are preset, the scanning area is set to 5m×5m; the initial grid size is set to 100mm; and the galvanometer focal length is set to 1500mm. For 1000 rad / s Voltage conversion coefficient , Continuing from 1.2 V / rad A fixed-density global scan was performed with a beam spacing of 0.5 mm to acquire the initial point cloud. Then, spatial mesh generation is performed, and the point cloud bounding box is calculated:

[0042] ;

[0043] ;

[0044] Units are in mm; via

[0045] , ,

[0046] Determine the number of grid cells for the X and Y axes.

[0047] This invention also discloses a method for calculating local feature parameters of the plurality of spatial grid cells, comprising: traversing the plurality of spatial grid cells to calculate the center point and determine the coordinates of the plurality of center points; extracting the coordinates of the plurality of points of the plurality of spatial grid cells and calculating the mean according to the three-dimensional coordinate axes to determine the centroid of the plurality of spatial grid cells; calculating the deviation between the plurality of point coordinates and the centroid to obtain the deviation vector and construct a covariance matrix; using the coordinates of the plurality of center points as the reference position to perform eigenvalue decomposition on the covariance matrix to obtain the plurality of eigenvalues ​​and normalizing them to obtain normalized eigenvalues; combining the normalized eigenvalues ​​with the parallel calculation of the plurality of spatial grid cells to obtain the information entropy of the plurality of spatial grid cells and generate the surface curvature of the plurality of spatial grid cells; and integrating the information entropy of the plurality of spatial grid cells with the surface curvature of the plurality of spatial grid cells to construct the local feature parameters.

[0048] Specifically, points are assigned to a grid, and the index of the grid cell containing a given point is calculated using the following formula. The continuous three-dimensional space is divided into regular grid cells, grid coordinates are assigned to each point cloud, and a corresponding spatial index is constructed so that each point cloud is assigned to the corresponding grid cell.

[0049] ;

[0050] in, It refers to the first i The coordinate values ​​on each coordinate axis i These are the X, Y, and Z axes, respectively. It is the size of the grid cell; It is a non-negative integer, starting from 0, representing the position of a point cloud at a given point. i The grid cell number corresponding to each coordinate axis direction is the point cloud's position in the [number of grid cells]. i Grid index on each coordinate axis; For point cloud bounding boxes in the first i The minimum coordinate on the axis, i.e., the lower bound of the distribution of all point clouds in that direction; through It can obtain the offset of the point cloud relative to the bounding box boundary, eliminating the influence of redundant values ​​of absolute coordinates.

[0051] Furthermore, the center position of each grid cell is calculated using the following formula, providing a reference position for local feature calculation and a corresponding starting center point for subsequent density decisions and helical scanning: ;in, Is a certain grid cell in the th i The geometric center coordinates on each coordinate axis are in the same units as the point cloud coordinates. Representing a point cloud in the th i The grid cell number corresponding to each coordinate axis direction; For point cloud bounding boxes in the firsti The minimum coordinate on the axis, that is, the lower limit of the distribution of all point clouds in that direction; It is the size of the grid cell.

[0052] Furthermore, local information entropy is used to characterize the geometric complexity of the local surface of the point cloud. This method constructs a covariance matrix to describe the spatial distribution characteristics of the local point cloud. By calculating the covariance matrix and then performing eigenvalue decomposition, three eigenvalues ​​are obtained. These eigenvalues ​​reflect the variance of the point cloud in the three principal directions.

[0053] The covariance matrix is ​​a 3×3 symmetric matrix whose elements are the pairwise covariances of the point cloud along the X, Y, and Z coordinate axes. Then, eigenvalue decomposition is performed on the covariance matrix to obtain three eigenvalues. , , The three eigenvalues ​​are usually sorted from smallest to largest. Then, the three eigenvalues ​​are normalized to obtain normalized eigenvalues, which are the variance contribution ratios in each principal direction. Information entropy calculation uses the normalized eigenvalues ​​to calculate Shannon entropy. The higher the entropy value, the more uniform the point cloud distribution, i.e., isotropic and complex surface. The lower the entropy value, the more concentrated the point cloud distribution, i.e., anisotropic and flat surface.

[0054] Furthermore, for the point cloud within each grid cell, calculate the mean of each coordinate component (X, Y, Z). Calculate the deviation of each point's coordinates from the mean. Calculate the covariance matrix using the deviation vector. First, set a grid with n points... The coordinates are the indices of the points, where the coordinates of each point are... To obtain all points within a grid cell: ;in, P This represents the set of all point clouds within a given grid cell. n This represents the number of point clouds; , Indicates the first A column vector of three-dimensional coordinates of a point cloud. , , These are the X, Y, and Z axis coordinates, respectively, with superscripts. This represents the transpose, which facilitates vector operations; subsequently, the formula is used... , Calculate the centroid of the point cloud. This represents the centroid of the point cloud, i.e., the average coordinates. The average value of the X-axis coordinate. The average value of the Y-axis coordinate. The average value of the Z-axis coordinates; the centroid is the reference point for measuring the spatial distribution of the point cloud, reflecting the overall average position.

[0055] Then through the formula

[0056] Calculate the deviation vector; where, For the first The deviation vector of a point relative to the centroid reflects the degree of deviation of a single point from the overall average position and is the basis for quantifying discreteness.

[0057] Then, a covariance matrix is ​​constructed based on the deviation vector. C for:

[0058] ;

[0059] Other, , and Let a and b represent the mean values ​​of the coordinate components a and b, respectively. Then, perform eigenvalue decomposition on the covariance matrix and solve for the characteristic equation:

[0060] ;

[0061] The eigenvalues ​​of the calculated covariance matrix , , And sorted in ascending order, i.e. . The value is used to characterize surface properties. The larger the value, the smaller the variation along the normal direction, and the flatter the surface. The smaller the value, the more concentrated the point cloud is in the normal direction, and the surface may have bumps, depressions or cracks.

[0062] Then the eigenvalues ​​are normalized using the formula. We obtain the normalized eigenvalues, where, Indicates the first j There are 1 normalized eigenvalues, with values ​​ranging from [0,1]. Corresponding to the original feature values, j =0,1,2, arranged in ascending order, are the three original values ​​obtained after the eigenvalue decomposition of the covariance matrix, reflecting the degree of dispersion of the point cloud in the three principal directions.

[0063] Subsequently, based on the normalized eigenvalues ​​and multiple spatial grid cells, parallel computation is performed to obtain the information entropy and surface curvature of multiple spatial grid cells. Parallel computation treats each spatial grid cell as an independent processing unit. The two computational tasks—calculating information entropy based on normalized eigenvalues ​​and calculating surface curvature based on normalized eigenvalues—are split into multiple parallel subtasks along the grid cell dimension. These subtasks are distributed to multiple computing cores for synchronous execution using a shared memory model or message passing model, avoiding the efficiency bottleneck of serial computation. Specifically, information entropy is calculated using the formula... The calculation shows that a larger value indicates a more dispersed distribution of normal vectors, meaning a more complex surface that requires higher density scanning. Information entropy is expressed in bits, and its value ranges from 1 to 10. ; Used to measure the information contribution of a single feature value, +10 -10 avoid The logarithm is meaningless; information entropy H It is an indicator for measuring the uniformity of point cloud distribution; the negative sign ensures... H A positive value indicates a more balanced dispersion in the three principal directions, a more uniform point cloud distribution, and a more complex surface structure. For example, the smaller the difference between the three normalized eigenvalues, the more uniform the dispersion in the three principal directions. , H The closer to the maximum value, the more balanced the dispersion of the point cloud in the three principal directions, and the more complex the surface structure, such as a region with dense cracks; when a certain feature value has an extremely high proportion, such as , , H A value close to 0 indicates that the point cloud is mainly distributed along a single direction, with a simple surface structure, such as the smooth surface of a complete rock mass. The curvature formula is derived through... The calculation reflects the degree of curvature of the surface along the normal direction. The principal curvature direction is the principal direction of the surface corresponding to the eigenvector of the covariance matrix; the curvature intensity is the magnitude of the eigenvalue reflecting the degree of curvature of the surface; the normal vector is the eigenvector corresponding to the smallest eigenvalue, which is the local normal direction.

[0064] For example, using a grid (25, 30), calculate the grid center:

[0065] ;

[0066] .

[0067] Subsequently, local feature calculations are performed, and the centroid is calculated by taking the point cloud within the grid. Construct the covariance matrix C :

[0068] ;

[0069] Eigenvalue decomposition yields: After sorting, it becomes The normalized eigenvalues ​​are:

[0070] ;

[0071] ;

[0072] ;

[0073] The information entropy and curvature are calculated as follows:

[0074] ;

[0075] .

[0076] The present invention also discloses a method for identifying the region type of multiple spatial grid cells based on the local feature parameters, and dynamically setting the laser beam density and beam spacing based on the identification results. The method includes: identifying the region type of multiple spatial grid cells based on the local feature parameters, generating an identification result, wherein the identification result includes multiple region types; traversing multiple spatial grid cells for matching based on the multiple region types, generating a matching result; and dynamically setting the laser beam density and beam spacing of the multiple spatial grid cells according to the matching result.

[0077] Specifically, the process of identifying region types and dynamically setting laser parameters based on local feature parameters relies on information entropy, curvature, and quality factors to construct a judgment logic. First, local feature parameters of grid cells are classified using preset thresholds, resulting in identification results encompassing three region types. Next, all spatial grid cells are traversed based on the identification results, precisely matching each grid with its corresponding region type to ensure each grid has a clear type classification, generating matching results. Finally, parameters are dynamically allocated according to the matching results, achieving precise adaptation between scanning parameters and region features.

[0078] like Figure 2 ,like Figure 3 This invention also discloses a method for identifying the region types of multiple spatial grid cells based on the local feature parameters, generating identification results, wherein the identification results include multiple region types. The method includes: setting a curvature threshold based on the surface curvature of the multiple spatial grid cells; performing extreme value analysis based on the information entropy of the multiple spatial grid cells to set a first information entropy threshold and a second information entropy threshold; comparing the surface curvature of the multiple spatial grid cells with the curvature threshold and the information entropy of the multiple spatial grid cells with the first information entropy threshold; and determining whether the surface curvature of any one of the multiple spatial grid cells is greater than the curvature threshold or the information entropy of any one of the multiple spatial grid cells is greater than the curvature threshold. If the information entropy is greater than the first information entropy threshold, then the first grid cell is extracted as the edge region of the structural surface; a quality factor evaluation is performed based on multiple spatial grid cells, and a first quality factor range and a second quality factor threshold are set; if the quality factor of any grid cell in the multiple spatial grid cells is within the first quality factor range, then the second grid cell is extracted as the fractured rock mass region; if the information entropy of any grid cell in the multiple spatial grid cells is less than the second information entropy threshold, then the third grid cell is extracted as the intact rock mass region; the edge region of the structural surface, the fractured rock mass region, and the intact rock mass region are categorized to determine the multiple region types.

[0079] Specifically, the original eigenvalues ​​are obtained by decomposing the eigenvalues ​​of the covariance matrix. Then, the eigenvalues ​​were normalized, and the information entropy was calculated separately. H With curvature k Next, the quality factor is calculated. Through formula The importance of a region is assessed comprehensively, with information entropy having a higher weight, as surface complexity better reflects the criticality of the geological structure.

[0080] Based on extensive references and practical engineering applications, a higher weight is assigned to information entropy. In actual geological structures, surface complexity reflects the importance of a region more comprehensively than local completeness. Therefore, the weights assigned to information entropy and curvature are 0.6 and 0.4, respectively. Based on the curvature value, information entropy, and quality factor calculated from the aforementioned local features, the scanned structural surface is classified into different regions according to different triggering conditions. Different laser beam densities and spacings are applied to different region types. For example, when curvature > 0.25 or entropy > 7.0, the region type is identified as a structural surface edge, and the laser beam density and beam spacing are 3.0 and 0.2 mm, respectively; when 0.6 ≤ quality factor ≤ 0.8, the region type is identified as fractured rock mass, and the laser beam density and beam spacing are 2.5 and 0.25 mm, respectively; when entropy < 4.0 or quality factor < 0.3, the region type is identified as intact rock mass, and the laser beam density and beam spacing are 0.8 and 0.8 mm, respectively.

[0081] For example, the quality factor is calculated as follows: When making dynamic density decisions, because Identified as a rock mass fracture region, density coefficient set to 2.5, beam spacing set to 0.25mm, target density... .

[0082] like Figure 4 This invention also discloses a method for analyzing regions in the identification results according to the laser beam density and the beam spacing to generate a spiral scanning path. The method includes: identifying regions according to the multiple region types to determine the region to be scanned; mapping the region to be scanned to multiple spatial grid cells to determine multiple target grid cells; extracting the center points of the multiple target grid cells as starting center points; performing spiral scanning analysis based on the density of the multiple target points of the multiple target grid cells to generate spiral path parameters; and performing a dense scan of the region to be scanned from the starting center point in combination with the spiral path parameters based on the laser beam density and the beam spacing to generate the spiral scanning path.

[0083] Specifically, region identification is performed according to the multiple region types. The region to be scanned is identified based on three previously identified regions: structural surface edge regions, fractured rock mass regions, and intact rock mass regions. The region to be scanned is the rock mass in the structural surface edge region. Structural surface edges are crucial for tunnel stability analysis, but traditional raster scanning is prone to undersampling in these regions. Therefore, this invention designs an adaptive spiral scanning method for enhanced scanning of structural surface edges. The spiral path is used to perform denser scanning in the identified edge regions. The core idea is to start from the center point of a single grid cell and scan outwards in a spiral pattern, with the radius gradually decreasing, thereby obtaining a higher point density in the central region. An adaptive scanning path is designed specifically for the structural surface edge region to enhance scanning accuracy. First, the path parameters are determined, including the coordinates of the grid cell center. Target point density Initial radius of the spiral , Set the grid size to 50% of the grid size; minimum radius. Stop when the radius is less than this value; the total number of turns of the spiral is N; then calculate the attenuation coefficient k, requiring that when When the radius decays to The requirement is to use the formula. Solve; then discretize the number of spiral turns, and assume the total number of turns. Calculate the current radius of each ring. ;in, , This represents the angle corresponding to the i-th ring, in radians, which increases linearly with the number of rings to ensure uniform coverage from 0 to 2. πN The total angle; This indicates that the radius decreases exponentially with increasing angle, causing the spiral trajectory to change from... Gradually shrink to Current number of ring points and angle step size ; among them, 2 πr i This represents the perimeter, or radius, of the i-th ring. r i The circumference of the circle; Indicates the number of points per unit length; Ensure that each ring has at least 5 points to avoid having too few points due to an excessively small radius, and to guarantee the uniform distribution of points on the ring. This represents the angular difference (in radians) between two adjacent points on the i-th ring, ensuring a uniform distribution of points on the ring. A smaller step size results in a denser point distribution. The polar coordinates are then converted to three-dimensional Cartesian coordinates. Calculate the coordinates of the point Connecting the points in the path list in sequence yields the spiral scan path; where, Indicates the i-th ring. j The angle at each point, in radians, is obtained by accumulating the angles using angle increments. j From 0 to Ensure that the points are evenly distributed along the ring; and polar coordinates The projection along the X and Y axes is obtained by referring to the coordinates of the grid center. By superimposing, the polar coordinates are converted into three-dimensional Cartesian coordinates, so that the scanning points form a spiral trajectory in the grid plane; This means the Z-axis coordinate of the scan point is always equal to the Z-coordinate of the grid center, ensuring the spiral path lies within the same height plane, suitable for scanning planar or near-planar regions such as structural edges. Finally, the formula is used... Minimize the total movement path length to ensure efficient path execution; among which, For scan point index, The coordinates of the scan point; Indicates the first The scan point and the first Euclidean distance between scan points; This represents the total distance between all consecutive scan points, i.e., the total path length the galvanometer needs to move; min indicates minimizing the total path length by adjusting the order of the scan points, such as using a natural spiral order instead of a random order. This optimization reduces unnecessary galvanometer movement, lowers scan time, and avoids vibration errors caused by excessively long paths, thus improving scan accuracy.

[0084] For example, a spiral scan path is generated, and a center point is set. initial radius ;Minimum radius Total number of laps N =4; Number of discrete segments K =16 segments / turn; Calculate the attenuation coefficient and the number of discretized spiral turns:

[0085] ,

[0086] Total number of rings is Ring; taking the 8th ring as an example, i =8, ; ; ; ; ; ; Calculate the coordinates of the points, convert them to Cartesian coordinates, connect the path points, and connect each point in sequence to obtain the spiral path.

[0087] like Figure 5This invention also discloses a method for converting the helical scanning path into a galvanometer control signal, and controlling the laser beam to scan along the helical scanning path using the galvanometer control signal. The method includes: traversing all path points within the scanning path and marking the points to determine multiple path points, wherein the scanning path includes a helical scanning path generated for the edge of an identified structural surface; extracting the three-dimensional Cartesian coordinates of the multiple path points and converting them to generate a galvanometer deflection angle; constructing a galvanometer control signal based on the galvanometer deflection angle; controlling the laser beam to traverse the scanning points of the scanning path using the galvanometer control signal to obtain a scanning result, wherein the scanning result includes a control signal sequence; and precisely controlling the laser beam density and the beam spacing based on the control signal sequence.

[0088] Specifically, all path points within the scanning path are traversed and marked to identify multiple path points. The scanning path includes a spiral scanning path generated for the edges of the identified structural surfaces. The galvanometer control execution system converts the path points in the Cartesian coordinate system into control signals for the galvanometer, i.e., voltage signals, and sends them to the galvanometer for execution according to a time sequence. First, a coordinate-angle transformation is performed, using the origin of the galvanometer system as a reference. The laser travels along the Z-axis, using the formula... and ,in, These are the coordinates of the galvanometer center on the scanning plane. f Convert the path coordinates to the galvanometer deflection angle for the focal length; , These are the deflection angles of the galvanometer in the X and Y axes, respectively, and are direct parameters controlling the rotation of the galvanometer. By deflecting the galvanometer at specific angles, the propagation direction of the laser beam can be changed, allowing the laser spot to accurately fall on the target point on the scanning path; x , y This represents the two-dimensional coordinates of a point on the scanning path in a plane perpendicular to the laser propagation direction, typically in millimeters. It originates from the spiral scanning path or scanning path data from other regions generated earlier. x 0, y 0) represents the projected coordinates of the galvanometer center on the scanning plane, in millimeters; that is, the point where the laser spot lands on the scanning plane when the galvanometer is not deflected. This parameter is the reference point for coordinate transformation, obtained through... , The offset of the target point relative to the reference point can be obtained, ensuring the relativity of the angle calculation.

[0089] Next, angle-to-voltage conversion is performed using a linear formula, which is:

[0090] In the formula: , These are the voltage signals (V) that drive the galvanometer to deflect in the X and Y axes, respectively. , Voltage-angle conversion factor , representing the driving voltage required per unit angle; , The zero-point offset voltage (V) converts the angle into a control voltage (V); , These are the deflection angles of the galvanometer in the X and Y axes, respectively, in radians.

[0091] Then, the timing sequence is arranged, and the start time is initialized. Calculate the adjacent coordinate offsets , Thus, the change in angle is obtained. ; ;Pick Through formula Calculate the time step to ensure the galvanometer does not exceed its speed limit; among which, This represents the change in angle, expressed in radians. and These are the X and Y axis deflection angles of the galvanometer corresponding to the current point. and It is the galvanometer deflection angle corresponding to the previous scan point. and By taking the absolute value of the angle difference, we can obtain the amount of angle change of the galvanometer from the previous position to the current position in the X and Y axes (in radians), which reflects the range of rotation required by the galvanometer. It is the maximum value among the changes in X and Y axis angles, used to simplify time step calculations, and to ensure motion safety by using the axis with the largest rotation amplitude of the galvanometer as the reference. The magnitude of the angular change of the galvanometer between two adjacent points, in radians; The maximum angular velocity of the galvanometer is expressed in radians per second. The shortest movement time based on velocity constraints, i.e., when the galvanometer rotates at its maximum angular velocity, is required to complete the angular change. Theoretical time required; 5 μs The minimum time step set for the system; To get 5 μs and The larger value; By taking 5 μs and The maximum of the two values ​​determines the time step.

[0092] pass A precise timeline is established, marking the laser trigger moment. Finally, a control signal is generated, including time, X-axis angle, Y-axis angle, and the laser trigger signal. The system sends these signals to the galvanometer and laser in a time sequence to activate the laser and complete the scan. This indicates the time step between two adjacent scan points, in microseconds, which is the time required for the galvanometer to move from the current point to the next point; For scan point index, Indicates the first The timestamp of each scan point is in microseconds, which is the laser trigger time corresponding to that point; Indicates the first The timestamp of each scan point is the preceding reference of the time series.

[0093] For example, galvanometer control is performed at (2573, 3050, 0), assuming the previous point coordinates are (2572.8, 3049.9, 0), and the previous point angle is... , ; Calculate coordinate offset: ; ;

[0094] The angle conversion and angle change calculation are as follows:

[0095] ;

[0096] ;

[0097] ; ; The calculation time step is: ; Convert to control voltage: , ; , , At this point, the laser beam is triggered according to the timestamp and control voltage of each point. After allocating the timestamp, the control signal is sent to trigger the laser to complete the scan.

[0098] This invention also discloses a method for precisely controlling the laser beam density and the beam spacing based on the control signal sequence. The method includes: setting a desired point cloud density threshold; extracting point cloud scanning data based on the control signal sequence; comparing the point cloud scanning data with the desired point cloud density threshold; if there is a deviation between the point cloud scanning data and the desired point cloud density threshold, calculating the deviation data; adaptively adjusting the laser beam density and the beam spacing according to the deviation data to generate an adjustment result; and updating the control signal sequence according to the adjustment result to generate an optimized control signal sequence.

[0099] Specifically, precise control of laser beam density and beam spacing based on the control signal sequence requires a four-step closed loop: threshold setting, data comparison, deviation adjustment, and signal optimization. First, setting the desired point cloud density threshold needs to be tailored to the specific region type; for the edge regions of structural surfaces, where the highest precision is required, the threshold is set to 16 points / ... mm 2 The corresponding beam density is 3.0 and the spacing is 0.2 mm; the fractured rock mass area is set to 10 points / mm 2 The beam density is 2.5 and the spacing is 0.25 mm; the intact rock mass area is set to 0.64 points / ... mm 2 The beam density is 0.8 and the spacing is 0.8 mm to ensure that the threshold matches the importance of the region. Secondly, actual point cloud scanning data is extracted based on the control signal sequence: from the sequence containing timestamps, galvanometer X / Y axis voltages, and laser trigger signals, the actual number of scanned points per unit area is calculated by the ratio of the number of path points to the corresponding grid area. For example, the actual scanned grid at the edge of a certain structural surface is 12 points / ... mm 2 Then, the actual data is compared with the expected threshold for the corresponding region.

[0100] If a discrepancy is found during comparison, such as the actual 12 points at the edge of the structural surface... mm 2 Below 16 points / mm 2 Or, the actual 1.2 points of the complete rock mass. mm 2 Above 0.64 points / mm 2 Then calculate the deviation data, the absolute deviation, such as , or relative deviation, such as The parameters are adaptively adjusted based on the deviation value. When the density is insufficient, the laser beam density is increased proportionally to the deviation. For example, if the deviation of the structural surface edge is 25%, the density is increased from 3.0 to 3.75, and the beam spacing is reduced from 0.2 mm to 0.16 mm. When the density is redundant, the density is reduced. If the deviation of the intact rock mass is 87.5%, the density is reduced from 0.8 to 0.43, and the beam spacing is increased from 0.8 mm to 1.5 mm, generating the adjustment result.

[0101] Finally, the control signal sequence is updated according to the adjustment results, the step size of the galvanometer angle is corrected, and the step size is reduced when the spacing is reduced; the laser trigger frequency is increased when the density is increased, and the optimized control signal sequence is generated to ensure that the point cloud density of subsequent scans accurately matches the expected threshold, thereby achieving closed-loop optimization of dynamic control.

[0102] This invention discloses a method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer adjustment. The method includes: performing an initial three-dimensional laser scan of a target area to acquire initial point cloud data; dividing the initial point cloud data into multiple spatial grid units and calculating local feature parameters for each spatial grid unit; identifying the region type of each spatial grid unit based on the local feature parameters, and dynamically setting the laser beam density and beam spacing based on the identification results; analyzing the regions in the identification results according to the laser beam density and beam spacing to generate a spiral scanning path; converting the spiral scanning path into a galvanometer control signal, controlling the laser beam to scan along the spiral scanning path using the galvanometer control signal, and precisely controlling the laser beam density and beam spacing based on the scanning results. This method solves the technical problem that traditional three-dimensional laser scanning cannot adjust the spatial distribution density of the beam in real time, making it difficult to meet the requirements for complete scanning of differentiated precision structural surfaces in various scenarios. It achieves precise control of the beam density in three-dimensional laser scanning, improving scanning accuracy and efficiency while reducing redundant data, thus meeting the scanning needs of scenarios such as tunnel surrounding rock and geological structures.

[0103] Compared with traditional solid-state scanning methods, this method significantly increases the point density in structural edge regions and fractured rock mass regions, thereby increasing the effective data volume in key areas while slightly reducing the data volume in non-critical areas, significantly reducing storage requirements and subsequent processing time. Furthermore, traditional scanning methods rely on manually preset scanning parameters, which cannot adapt to complex geological structures. This invention achieves fully adaptive decision-making throughout the entire process.

[0104] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0105] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for precise control of beam density in three-dimensional laser scanning based on dynamic galvanometer adjustment, characterized in that, The method includes: Perform an initial 3D laser scan of the target area to obtain initial point cloud data; The initial point cloud data is divided into multiple spatial grid cells, and the local feature parameters of the multiple spatial grid cells are calculated. Based on the local feature parameters, the region type of multiple spatial grid cells is identified, and the laser beam density and beam spacing are dynamically set based on the identification results. The region in the recognition result is analyzed according to the laser beam density and the beam spacing to generate a spiral scanning path; The spiral scanning path is converted into a galvanometer control signal, and the laser beam is controlled to scan along the spiral scanning path by the galvanometer control signal. The laser beam density and the beam spacing are precisely adjusted according to the scanning results. The region in the recognition result is analyzed according to the laser beam density and the beam spacing to generate a spiral scanning path. The method includes: Region identification is performed according to the multiple region types to determine the region to be scanned; The area to be scanned is mapped to multiple spatial grid cells to determine multiple target grid cells; Extract the center points of the multiple target grid cells as the starting center points; Helical scan analysis is performed based on the density of multiple target points in the multiple target grid cells to generate helical path parameters; Based on the laser beam density and the beam spacing, the area to be scanned is densified from the starting center point in combination with the spiral path parameters to generate the spiral scanning path; Region identification is performed according to the multiple region types mentioned above. The region to be scanned is identified according to three types of regions: structural plane edge region, fractured rock mass region, and intact rock mass region. Among them, the region to be scanned is the rock mass of the structural plane edge region. An adaptive spiral scanning method is used for enhanced scanning of structural plane edges. The spiral path is used to perform densified scanning in the identified edge region. The core idea is to start from the center point of a unit grid and scan outward in a spiral form, with the radius gradually decreasing, so as to obtain a higher point density in the central region. The method for calculating the local characteristic parameters of the plurality of spatial grid cells includes: The center point is calculated by traversing multiple spatial grid cells to determine the coordinates of multiple center points; The coordinates of multiple points in multiple spatial grid cells are extracted and averaged according to the three-dimensional coordinate axes to determine the centroid of multiple spatial grid cells; Based on the coordinates of the multiple points and the centroid, the deviation is calculated to obtain the deviation vector and construct the covariance matrix. Using the coordinates of the multiple center points as reference positions, the covariance matrix is ​​decomposed into eigenvalues ​​to obtain multiple eigenvalues. These eigenvalues ​​are then normalized to obtain normalized eigenvalues. Based on the normalized eigenvalues, parallel computation is performed using multiple spatial grid cells to obtain the information entropy and surface curvature of multiple spatial grid cells. The information entropy of multiple spatial grid cells is integrated with the surface curvature of multiple spatial grid cells to construct the local feature parameters; Based on the local feature parameters, the method identifies the region types of multiple spatial grid cells and generates an identification result, wherein the identification result contains multiple region types. A curvature threshold is set based on the surface curvature of multiple spatial grid cells, and an extreme value analysis is performed based on the information entropy of multiple spatial grid cells to set a first information entropy threshold and a second information entropy threshold. The surface curvature of multiple spatial grid cells is compared with the curvature threshold, and the information entropy of multiple spatial grid cells is compared with the first information entropy threshold to determine the result. If the surface curvature of any one of the multiple spatial grid cells is greater than the curvature threshold or the information entropy of any one of the multiple spatial grid cells is greater than the first information entropy threshold, then the first grid cell is extracted as the edge region of the structural surface. Quality factor evaluation is performed based on multiple spatial grid cells, and a first quality factor interval and a second quality factor threshold are set. If the quality factor of any one of the multiple spatial grid cells is within the first quality factor range, then the second grid cell is extracted as the fractured rock mass region. If the information entropy of any one of the multiple spatial grid cells is less than the second information entropy threshold, then the third grid cell is extracted as the complete rock mass region. The edge region of the structural surface, the fractured rock mass region, and the intact rock mass region are categorized and identified to determine the types of the multiple regions.

2. The method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer control as described in claim 1, characterized in that, Dividing the initial point cloud data into multiple spatial grid units includes the following methods: The initial point cloud data is marked with extreme values ​​according to the three-dimensional coordinate axes to determine the spatial range of the point cloud. Set the grid cell size, divide the three-dimensional coordinate axes according to the grid cell size, and determine the number of grids; Using the spatial range of the point cloud as a constraint, the initial point cloud data is divided according to the number of grids to generate the multiple spatial grid units.

3. The method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer control as described in claim 1, characterized in that, The method involves identifying the region type of multiple spatial grid cells based on the local feature parameters, and dynamically setting the laser beam density and beam spacing based on the identification results. Based on the local feature parameters, the region types of multiple spatial grid cells are identified, and an identification result is generated, which includes multiple region types. Based on the multiple region types, multiple spatial grid cells are traversed for matching to generate matching results; The laser beam density and beam spacing are dynamically set for multiple spatial grid cells according to the matching results.

4. The method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer control as described in claim 1, characterized in that, The method involves converting the helical scanning path into a galvanometer control signal, and then controlling the laser beam to scan along the helical scanning path using the galvanometer control signal. All path points within the scanning path are traversed and marked to identify multiple path points. The scanning path includes a spiral scanning path generated for the edges of the identified structural surfaces. The three-dimensional Cartesian coordinates of the multiple path points are extracted and transformed to generate the galvanometer deflection angle. The galvanometer control signal is then constructed based on the galvanometer deflection angle. The laser beam is controlled by the galvanometer control signal to scan the scanning points of the scanning path to obtain the scanning result, which includes a sequence of control signals. The laser beam density and the beam spacing are precisely controlled based on the control signal sequence.

5. The method for precise control of three-dimensional laser scanning beam density based on dynamic galvanometer control as described in claim 4, characterized in that, The method for precisely controlling the laser beam density and the beam spacing based on the control signal sequence includes: Set a desired point cloud density threshold, extract point cloud scanning data based on the control signal sequence, and compare the point cloud scanning data with the desired point cloud density threshold; If the point cloud scanning data deviates from the desired point cloud density threshold, the deviation data is calculated, and the laser beam density and the beam spacing are adaptively adjusted based on the deviation data to generate an adjustment result. The control signal sequence is updated according to the adjustment results to generate an optimized control signal sequence.

Images