A method for identifying dangerous rock masses in high-steep mountainous areas

Abstract

The present invention discloses a method for identifying dangerous rock masses in high-steep mountainous areas, which relates to the technical field of dangerous rock mass identification. The method includes: collecting first image data of a target area based on a first collection device, and constructing a roughness model based on the first image data; obtaining flight route parameters based on the first image data, collecting second image data of the target area based on the flight route parameters and a second collection device, and constructing a three-dimensional image model and a point cloud model based on the second image data and the roughness model; obtaining dominant structural planes based on the three-dimensional image model and the point cloud model, and obtaining structural plane parameters based on the dominant structural planes; identifying dangerous rock masses based on the structural plane parameters and a collapse stability judgment table, which can solve the problems of high difficulty, high risk and low efficiency in manually surveying dangerous rock masses in high-steep mountainous areas, and low identification accuracy in the UAV survey method.

Description

Technical Field

[0001] The present invention relates to the technical field of dangerous rock mass identification, and specifically, to a method for identifying dangerous rock masses in high-steep mountainous areas. Background Art

[0002] Comprehensive investigation and identification as well as advance protection are the most effective methods for preventing and controlling dangerous rock masses. However, factors such as the complex geological environment background conditions in mountainous areas, superposition of earthquakes and rainfall, and backward disaster investigation means make it difficult to conduct geological disaster investigations, identifications, and evaluations in mountainous areas. Moreover, traditional geological investigation means, such as manual climbing investigations, are very difficult to conduct a comprehensive and detailed investigation of large-scale dangerous rock masses, especially in areas such as rock walls, high-steep slopes, residual slopes of landslides, severely fractured mountains, or areas where rockfalls frequently occur, and it is even more difficult for geological personnel to approach. Even when using drones to take pictures, due to the complex surface morphology of the slopes, the boundary lines of dangerous rock masses are not clear, and there is a risk of misjudging deformation.

[0003] Therefore, how to provide a method for identifying dangerous rock masses in high-steep mountainous areas is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0004] In order to solve the problems of high difficulty, high risk, and low efficiency in manually investigating dangerous rock masses in high-steep mountainous areas, and the low identification accuracy of the drone investigation method, the present invention provides a method for identifying dangerous rock masses in high-steep mountainous areas. The method includes: collecting first image data of a target area based on a first collection device, and constructing a roughness model based on the first image data; obtaining flight route parameters based on the first image data, collecting second image data of the target area based on the flight route parameters and a second collection device, and constructing a three-dimensional image model and a point cloud model based on the second image data and the roughness model; obtaining dominant structural planes based on the three-dimensional image model and the point cloud model, and obtaining structural plane parameters based on the dominant structural planes; and identifying dangerous rock masses based on the structural plane parameters and a collapse stability judgment table.

[0005] The dominant structural plane refers to: the structural plane that plays a controlling role in the stability of the rock mass found according to certain dominant indicators in the rock mass, and is the main structural plane that controls the deformation of rock slopes.

[0006] Principle of the present invention: First, use the first acquisition device to obtain image data of high-steep mountainous areas, establish a preliminary model based on the image data, and construct a more refined flight route and flight parameters according to the image data, so that the millimeter-level image data collected by the second acquisition device is more refined and clear, reducing the misjudgment of unclear dangerous rock mass boundary lines caused by unclear images. Based on the millimeter-level image data, construct a more accurate millimeter-level three-dimensional model and point cloud model on the initial model, making it closer to the actual terrain or object surface, and automatically identify the dominant structural planes based on the model. Then, automatically extract the structural plane parameters based on the dominant structural planes to obtain the main structural planes controlling the deformation of rock slopes, analyze their structural planes specifically, more effectively obtain the deformed dangerous rock masses, improve the recognition accuracy of dangerous rock masses, identify dangerous rock masses based on the structural plane parameters and the collapse stability judgment table, construct an identification system of data acquisition - model construction - automatic identification, improve the recognition accuracy of dangerous rock masses, and only need to manually operate the drone to collect data, with a faster data collection speed and higher efficiency. There is no need for manual climbing for investigation, reducing the investigation difficulty and the safety risk of investigators.

[0007] Further, the specific steps for obtaining the flight route parameters include: obtaining the first device parameters of the first acquisition device, obtaining the first parameters based on the first device parameters, where the first parameters include the ground coverage range and ground resolution, and the first device parameters include the focal length of the first acquisition device, the width and height of the sensor of the first acquisition device, the flight height of the first acquisition device, and the pixel size; obtaining the shooting interval between two adjacent images based on the first image data, obtaining the second parameters based on the shooting interval and the ground coverage range, where the second parameters include the forward overlap, the side overlap, and the flight line interval; obtaining the regional size of the target area, obtaining the third parameters based on the regional size, the shooting interval, and the heading interval, where the third parameters include the number of flight lines and the number of shooting points on each flight line; obtaining the flight route parameters based on the first parameters, the second parameters, and the third parameters.

[0008] The photo coverage range refers to the actual area on the ground that can be observed by each taken photo.

[0009] The shooting interval refers to the distance between two consecutive photos in the flight direction.

[0010] The forward overlap refers to the overlapping part between two consecutive photos.

[0011] The side overlap refers to the overlapping part between two adjacent flight lines on the same flight route.

[0012] The flight line interval refers to the interval between flight lines.

[0013] The ground resolution is the actual size on the ground corresponding to each pixel.

[0014] Furthermore, the specific steps for constructing the three-dimensional image model and the point cloud model include: obtaining the second device parameters of the second acquisition device, obtaining the interior orientation parameters based on the second device parameters, where the second device parameters include the focal length and the principal point coordinates; matching the second image data based on preset feature points to obtain corresponding points, and obtaining the exterior orientation parameters based on the corresponding points, where the exterior orientation parameters include the rotation matrix and the translation vector; obtaining the fourth parameter of the second image data, obtaining the three-dimensional space coordinates based on the corresponding points and the fourth parameter, where the fourth parameter includes the projection matrix and the scale factor; constructing a projection equation based on the interior orientation parameters, the exterior orientation parameters, and the three-dimensional space coordinates; constructing a triangular network based on the roughness model and the projection equation; obtaining the optimal texture image based on the second image data, and performing texture mapping on each triangular patch in the triangular network based on the optimal texture image to obtain the three-dimensional image model and the point cloud model. Selecting the texture image with the richest texture mapping information and ensuring the color consistency of the texture gaps between adjacent triangular patches from different texture images can ensure the accuracy of the three-dimensional model and provide high-quality data for subsequent extraction of structural plane parameters.

[0015] Furthermore, before constructing the triangular network, it also includes: optimizing the projection equation based on minimizing the reprojection error, and constructing the triangular network based on the roughness model and the optimized projection equation.

[0016] Furthermore, the specific steps for constructing the triangular network include: performing triangulation on the roughness model to obtain a triangular grid; dividing the triangular grid into several polygons to obtain Thiessen polygons; obtaining the normal vector based on preset triangle vertices; and performing surface reconstruction and surface fitting on the Thiessen polygons based on the normal vector and the projection equation to obtain the triangular network.

[0017] Further, the specific steps for obtaining the dominant structural plane include: acquiring the point clouds of the three-dimensional image model and the point cloud model, searching for the nearest neighbor points of the point clouds to obtain a set of nearest neighbor points; obtaining a covariance matrix based on the set of nearest neighbor points, and obtaining eigenvalues based on the covariance matrix; obtaining a deviation parameter based on the eigenvalues, and obtaining a first point cloud set based on the deviation parameter and the maximum deviation; obtaining the point normal vectors of the first point cloud set based on the eigenvalues; unifying the directions of the point normal vectors based on a preset direction to obtain a coplanar point cloud set; obtaining the dip angle and dip direction based on the coplanar point cloud set and the plane equation; obtaining the attitude of the structural plane based on the dip angle, the dip direction, and the point normal vectors; calculating the kernel density of the coplanar point cloud set to obtain the kernel density, dividing the first point cloud set based on the kernel density to obtain several density clusters, obtaining the maximum kernel density of each density cluster based on the kernel density, obtaining the density peak points based on the maximum kernel density, and performing clustering analysis on each density cluster based on the density peak points to obtain the dominant structural plane.

[0018] The dip angle refers to the angle between the steepest direction of the inclined line on the structural plane and the north direction; the dip direction refers to the maximum angle between the horizontal plane and the structural plane.

[0019] The attitude refers to the position and state of geological bodies (such as rock strata, ore bodies, etc.) in space. It describes the orientation and shape of geological bodies in space and is usually composed of three elements: strike, dip direction, and dip angle.

[0020] Further, the specific steps for performing clustering analysis on the density clusters based on the density peak points include:

[0021] S1. Arbitrarily select a region to obtain a seed region, and acquire the adjacent regions adjacent to the seed region;

[0022] S2. Calculate the grayscale means of the seed region and the adjacent regions respectively, and obtain the similarity based on the grayscale means;

[0023] S3. Determine whether the similarity is less than a preset value. If so, obtain the dominant structural plane based on the seed region and the adjacent regions; if not, merge the seed region and the adjacent regions to obtain a first region, return to S2, and update the seed region to the first region.

[0024] Further, the specific steps for obtaining the structural plane parameters include: based on a threshold and the attitude, fusing, correcting, and extracting the attitude information of the dominant structural plane in sequence to obtain the structural plane parameters.

[0025] Further, the specific steps for identifying dangerous rock masses include: obtaining the spatial distribution characteristics of the dominant structural planes, obtaining the first structural planes that intersect with each rock mass based on the spatial distribution characteristics, and obtaining the number of structural planes of the first structural planes; obtaining the attitude relationship between the first structural planes and the rock masses based on the structural plane parameters; obtaining the rock mass collapse information based on the number of structural planes, the attitude relationship, and the collapse stability judgment table; and identifying the dangerous rock masses based on the rock mass collapse information.

[0026] The spatial distribution characteristics refer to the extension of the structural planes in space, and are mainly reflected by three important parameters: strike, dip, and dip angle, which reflect their spatial distribution characteristics.

[0027] The intersection relationship refers to the spatial relationship between the structural planes related to the contact with the rock mass, between the structural planes, and between the structural planes and the rock mass.

[0028] The attitude relationship refers to the relationship between the attitude elements of the structural planes and the rock masses.

[0029] Further, the specific steps for identifying dangerous rock masses based on the rock mass collapse information include: obtaining the collapsed rock masses based on the rock collapse information, extracting the outcrop points of the collapsed rock masses; obtaining the second structural planes of the collapsed rock masses, and extracting the intersection lines of the structural planes of the second structural planes; obtaining the spatial coordinates and collapse numbers based on the outcrop points and the intersection lines of the structural planes, and marking the dangerous rock masses based on the spatial coordinates and the collapse numbers.

[0030] The outcrop point refers to the intersection point where the structural plane (such as fault, joint, bedding plane, etc.) is exposed on the surface of the rock mass, and these points are usually the intersection lines of the structural plane on the outer surface of the rock mass and the external environment.

[0031] The intersection line of the structural plane refers to the direction of the intersection line where the structural plane intersects with the horizontal plane.

[0032] One or more technical solutions provided by the present invention have at least the following technical effects or advantages:

[0033] 1. The first acquisition device is used to obtain the image data of the high-steep mountain area, a preliminary model is established based on the image data, and a more refined flight route and flight parameters are constructed according to the image data, so that the image data collected by the second acquisition device is more refined and clear, reducing the misjudgment caused by unclear images resulting in unclear boundary lines of dangerous rock masses.

[0034] 2. Construct a three-dimensional model and a point cloud model with a higher precision of millimeter level on the initial model to make it closer to the actual terrain or the surface of the object. Automatically identify the dominant structural planes based on the model, and then automatically extract the structural plane parameters based on the dominant structural planes to obtain the main structural planes that control the deformation of the rock slope. Analyze its structural planes specifically to more effectively obtain the dangerous rock masses with deformation, improve the recognition accuracy of dangerous rock masses, identify dangerous rock masses based on the structural plane parameters and the collapse stability judgment table, construct an identification system of data collection - model construction - automatic identification to improve the recognition accuracy of dangerous rock masses, and only need to manually operate the unmanned aerial vehicle to collect data, with a faster data collection speed and higher efficiency. There is no need for manual climbing and investigation, reducing the investigation difficulty and the safety risk of the investigators. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, form a part of the present invention, and do not limit the embodiments of the present invention;

[0036] Figure 1 is a schematic flow chart of a method for identifying dangerous rock masses in high-steep mountainous areas in the present invention;

[0037] Figure 2 is a schematic diagram of the aerial survey technical route in the present invention;

[0038] Figure 3 is a three-dimensional space schematic diagram of the structural plane and the interpretation parameters in the present invention;

[0039] Figure 4 is a schematic diagram of the three-dimensional collapse point cloud model in the present invention;

[0040] Figure 5 is a diagram of the automatically identified dominant structural plane in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] In order to more clearly understand the above objects, features and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0042] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0043] Embodiment 1

[0044] Refer to Figures 1-5 , Figure 3In it, ABCD represents the rock mass structural plane; α represents the dip direction, which is the inclination direction of the structural plane in space. It is the direction of the projection of a ray drawn vertically downward along the dip direction and perpendicular to the strike on the horizontal plane, and its magnitude is the angle between the horizontal projection line and the due north direction (0°), with the unit of degree (°); β represents the dip angle, which is the magnitude of the inclination angle of the structural plane in space and is represented by the acute angle between the structural plane and the horizontal plane, with the unit of degree (°).

[0045] This embodiment provides a method for identifying dangerous rock masses in high-steep mountainous areas. The method includes:

[0046] Collecting first image data of a target area based on a first collection device, and constructing a roughness model based on the first image data;

[0047] Obtaining flight route parameters based on the first image data, collecting second image data of the target area based on the flight route parameters and a second collection device, and constructing a three-dimensional image model and a point cloud model based on the second image data and the roughness model. In this embodiment, the first collection device can be a camera with a fixed wing or a multi-rotor carrying a single lens or a five-lens camera, and the second collection device can be a five-lens camera with a pan-tilt head.

[0048] Among them, the specific steps for obtaining the flight route parameters include: obtaining first device parameters of the first collection device, and obtaining first parameters based on the first device parameters. The first parameters include the ground coverage range and the ground resolution, and the first device parameters include the focal length of the first collection device, the width and height of the sensor of the first collection device, the flight height of the first collection device, and the pixel size;

[0049] In this embodiment, the first parameters can be obtained by a first calculation method. The first calculation method is:

[0050]

[0051] Among them, GSD w represents the width of the ground coverage range, GSD h represents the height of the ground coverage range, W represents the width of the sensor of the first collection device, H represents the height of the sensor of the first collection device, H f represents the flight height of the first collection device, f represents the focal length of the first collection device, GSD represents the ground resolution, and p represents the pixel size of the first collection device.

[0052] Obtain the shooting interval between two adjacent images based on the first image data, and obtain a second parameter based on the shooting interval and the ground coverage range. The second parameter includes the forward overlap degree, the lateral overlap degree, and the flight line interval; obtain the regional size of the target area, and obtain a third parameter based on the regional size, the shooting interval, and the heading interval. The third parameter includes the number of flight lines and the number of shooting points on each flight line; obtain the flight route parameters based on the first parameter, the second parameter, and the third parameter.

[0053] In this embodiment, a second calculation method can be used to obtain the second parameter. The second calculation method is as follows:

[0054]

[0055] S = GSD w ·(1 - O s ); (6)

[0056] Wherein, GSD w represents the width of the ground coverage range, GSD h represents the height of the ground coverage range, O f represents the forward overlap degree, O s represents the lateral overlap degree, S represents the flight line interval, and D represents the shooting interval. Preset the lateral overlap degree or the flight line interval, and calculate the other parameter according to formula (4) or formula (5).

[0057] In this embodiment, a third calculation method can be used to obtain the number of flight lines and the number of shooting points on each flight line. The third calculation method is as follows:

[0058]

[0059] Wherein, N p represents the number of shooting points on each flight line, N l represents the number of flight lines, L represents the length of the target area, W a represents the width of the target area, S represents the flight line interval, and D represents the shooting interval.

[0060] Among them, the specific steps for constructing the three-dimensional image model and the point cloud model include: obtaining the second device parameters of the second acquisition device, obtaining the internal orientation parameters based on the second device parameters, where the second device parameters include the focal length and the principal point coordinates; matching the second image data based on preset feature points to obtain corresponding points, such as using feature point matching algorithms, the SIFT algorithm (Scale-Invariant Feature Transform) or the SURF algorithm (Speeded Up Robust Features) to find the corresponding points in multiple images; obtaining the external orientation parameters based on the corresponding points, where the external orientation parameters include the rotation matrix and the translation vector, which respectively describe the position and orientation of the camera during shooting, and by matching the corresponding points in multiple images, the external orientation parameters can be calculated; obtaining the fourth parameter of the second image data, obtaining the three-dimensional space coordinates based on the corresponding points and the fourth parameter, where the fourth parameter includes the projection matrix and the scale factor; constructing a projection equation based on the internal orientation parameters, the external orientation parameters and the three-dimensional space coordinates; constructing a triangular network based on the roughness model and the projection equation; obtaining the optimal texture image based on the second image data, and performing texture mapping on each triangular patch in the triangular network based on the optimal texture image to obtain the three-dimensional image model and the point cloud model. In this embodiment, the preset feature points can be image control points, plane control points, elevation control points, etc.

[0061] In this embodiment, the internal orientation parameters can be obtained by using the fourth calculation method, and the fourth calculation method is:

[0062]

[0063] Among them, K represents the internal parameter matrix, f x and f y respectively represent the focal lengths in the x and y axis directions, c x and c y respectively represent the principal point coordinates in the x and y axis directions.

[0064] In this embodiment, the three-dimensional space coordinates can be obtained by using the fifth calculation method, and the fifth calculation method is:

[0065]

[0066] Among them, X represents the three-dimensional space coordinates, P1 and P2 respectively represent the projection matrices of two images, x1 and x2 respectively represent the coordinates of the corresponding points in the two images, and λ1 and λ2 both represent the scale factors.

[0067] In this embodiment, the projection equation can be obtained by using the sixth calculation method, and the sixth calculation method is:

[0068] x = K[R|t]X; (11)

[0069] Wherein, x represents the image coordinates, K represents the intrinsic parameter matrix, R represents the rotation matrix, X represents the three-dimensional space coordinates, and t represents the translation vector.

[0070] Wherein, before constructing the triangular mesh, it further includes: optimizing the projection equation based on minimizing the reprojection error, and constructing the triangular mesh based on the roughness model and the optimized projection equation. For example, the bundle adjustment method is used to optimize the positions of the three-dimensional points and the camera parameters, and the seventh calculation method is used for optimization. The seventh calculation method is as follows:

[0071]

[0072] Wherein, R represents the rotation matrix, X represents the three-dimensional space coordinates, t represents the translation vector, and x ij represents the observation value of the j-th three-dimensional point in the i-th camera, and X i represents the three-dimensional space coordinates of the i-th three-dimensional point, and R j represents the rotation matrix of the j-th camera, and t j represents the translation vector of the j-th camera, π represents the projection function, and both i and j represent serial numbers.

[0073] Wherein, the specific steps for constructing the triangular mesh include: using the Delaunay triangulation algorithm to triangulate the roughness model to obtain a triangular mesh, maximizing the minimum angle to avoid long and narrow triangles; using the Voronoi diagram to divide the triangular mesh into several polygons to obtain the Thiessen polygons; obtaining the normal vector based on the preset triangle vertices; based on the normal vector and the projection equation, performing surface reconstruction and surface fitting on the Thiessen polygons to obtain the triangular mesh. For example, the Poisson surface reconstruction method is used to perform surface reconstruction on the Thiessen polygons to generate a smooth surface, and the best surface fitting is obtained by minimizing the energy function, and it is calculated using the Poisson equation. The Poisson equation is as follows:

[0074]

[0075] Wherein, Δf represents the Laplacian operator, represents the divergence of the normal vector field. The normal vector n is obtained can be obtained by using the existing technology.

[0076] Then, the least squares fitting method is used to smooth the point cloud data to make the triangular mesh closer to the actual terrain or the surface of the object, so as to obtain the triangular mesh. Assuming the plane equation is ax1 + by1 + cz1 + d = 0, the calculation method of the least squares fitting method can be:

[0077]

[0078] Among them, a, b, c, and d all represent the coefficients to be solved, and x k , y k and z k all represent the three-dimensional point coordinates in space, m represents the number of three-dimensional point coordinates in space, k represents the serial number, and x1, y1, and z1 all represent coordinates.

[0079] In this embodiment, the triangular mesh can be further optimized. For example, optimization methods such as edge collapse, vertex translation, and topological transformation can be used to reduce the number of tumor-like triangles, smooth the surface, and improve the geometric and topological quality of the mesh.

[0080] In this embodiment, it is assumed that the preset triangle vertices are A(x1, y1, z1), B(x2, y2, z2), and C(x3, y3, z3) respectively. The normal vector can be obtained by using the eighth calculation method, and the eighth calculation method is as follows:

[0081] AB = (x2 - x1, y2 - y1, z2 - z1); (15)

[0082] AC = (x3 - x1, y3 - y1, z3 - z1); (16)

[0083] n = AB × AC = ((y2 - y1)(z3 - z1) - (z2 - z1)(y3 - y1), (z2 - z1)(x3 - x1) - (x2 - x1)(z3 - z1), (x2 - x1)(y3 - y1) - (y2 - y1)(x3 - x1)); (17)

[0084] Among them, n represents the normal vector, and AB and AC both represent the sides of the triangle.

[0085] Based on the three-dimensional image model and the point cloud model, the dominant structural plane is obtained, and the structural plane parameters are obtained based on the dominant structural plane;

[0086] Among them, the specific steps to obtain the dominant structural plane include: obtaining the point clouds of the three-dimensional image model and the point cloud model, searching for the nearest neighbor points of the point clouds to obtain the nearest neighbor point set. For example, using a K-D tree (K-Dimensional Tree) as an index to search for the nearest neighbor points of the point clouds, by dividing the spatial data layer by layer, the nearest neighbor point set of each point can be found efficiently; obtaining the covariance matrix based on the nearest neighbor point set, and obtaining the eigenvalues based on the covariance matrix; obtaining the deviation parameter based on the eigenvalues, and obtaining the first point cloud set based on the deviation parameter and the maximum deviation; obtaining the point normal vector of the first point cloud set based on the eigenvalues. For example, using the PCA (Principal Component Analysis) algorithm to extract the point cloud normal vector, this method is a dimensionality reduction analysis method, which can obtain the main direction of the point cloud through the eigenvalue decomposition of the covariance matrix. The calculation formula for obtaining the covariance matrix is:

[0087]

[0088] Among them, COV represents the covariance matrix, P i represents the i-th point cloud, represents the centroid of the point cloud, T represents the transpose, u represents the number of point clouds, and i represents the serial number.

[0089] The calculation formula for obtaining the eigenvalues is:

[0090] COV = W1∧W1 T ;(19)

[0091] Among them, COV represents the covariance matrix, W1 represents the eigenvector matrix, T represents the transpose, and Λ represents the diagonal matrix. According to this formula, the diagonal matrix is obtained, and its diagonal elements are the eigenvalues, which can be respectively denoted as and

[0092] Using the eigenvalues obtained by the PCA algorithm for coplanarity test, the coplanarity of the point set is judged by calculating the deviation parameter. The calculation of the deviation parameter can be:

[0093]

[0094] Among them, and both represent eigenvalues, and η represents the deviation parameter. Define the parameter tolerance η max is defined as the maximum allowable deviation in the point subset. If η > η max , then this subset is discarded.

[0095] The three eigenvalues correspond to the directions of the new coordinate system. The eigenvalues and reflect the extensibility of the point cloud in each direction. For the local point cloud on a plane, and much greater than the minimum eigenvalue The corresponding eigenvector is the normal vector.

[0096] Based on the preset direction, unify the directions of the point normal vectors to obtain a coplanar point cloud set; for example, if the z component C of the normal vector is positive, perform coordinate transformation on the normal vectors in the opposite direction. If C < 0, the new normal vector is n1 = (-A, -B, -C); if C ≥ 0, then n1 = (A, B, C).

[0097] Obtain the dip angle and dip direction based on the coplanar point cloud set and the plane equation; obtain the attitude of the structural plane based on the dip angle, the dip direction, and the point normal vector; for example, set the plane equation as:

[0098] Ax2 + By2 + Cz2 + D = 0; (21)

[0099] Calculate the dip angle and dip direction according to this formula:

[0100]

[0101] where A, B, and C all represent plane coefficients, D represents the perpendicular distance from the origin to the plane, β represents the dip angle, α represents the dip direction, α0 represents a parameter, and x, y, and z all represent coordinates.

[0102] Calculate the point cloud density of the coplanar point cloud set to obtain the kernel density, divide the first point cloud set based on the kernel density to obtain several density clusters, obtain the maximum kernel density of each density cluster based on the kernel density, obtain the density peak points based on the maximum kernel density, and perform clustering analysis on each density cluster based on the density peak points to obtain the dominant structural plane. For example, use the kernel density estimation method to calculate the point cloud density, process the scattered point cloud through the Gaussian kernel density estimation function, and perform density division to obtain the density peak points as the clustering centers. The calculation method adopted by the kernel density estimation method is:

[0103]

[0104] The calculation method adopted by the Gaussian kernel density estimation function is:

[0105]

[0106] where f KDE (ξ) represents the kernel density estimation function, K σ (ξ, ξ1) represents the Gaussian kernel density estimation function, ξ1, …, ξ n1 , ξ i ∈R d, n1 represents a set of random observations of the probability density function, i represents the serial number, both ξ and ξ1 represent two-dimensional vectors, d1 represents the dimension of the data. Since both ξ and ξ1 are two-dimensional vectors, d1 = 2, σ represents the kernel function width, K1 represents the kernel function, satisfying K1 ≥ 0 and ∫K(ξ,·)dξ = 1.

[0107] Among them, the watershed clustering algorithm can be adopted to perform clustering analysis on the density clusters based on the density peak points. However, it will cause over-segmentation. Therefore, in this embodiment, the watershed algorithm is improved by merging adjacent regions to solve the above problems. The specific steps include:

[0108] S1. Arbitrarily select a region to obtain a seed region, and obtain adjacent regions adjacent to the seed region;

[0109] S2. Calculate the gray mean values of the seed region and the adjacent regions respectively, and obtain the similarity based on the gray mean values. The calculation method can be:

[0110]

[0111] Among them, S1 represents the similarity, S i and S t1 both represent the gray mean values of the adjacent regions, T1 represents the empirical value, n2 represents the number of adjacent regions, and both t1 and i represent the serial numbers;

[0112] S3. Judge whether the similarity is less than the preset value. If so, obtain the dominant structural plane based on the seed region and the adjacent regions; if not, merge the seed region and the adjacent regions to obtain a first region, return to S2, and update the seed region to the first region.

[0113] Among them, the specific steps for obtaining the structural plane parameters include: based on the threshold and the attitude, fuse, correct, and extract the attitude information of the dominant structural plane in sequence to obtain the structural plane parameters. For example, based on the similarity in step S2 above, set the similarity threshold, divide the dominant structural plane, obtain the dominant structural groups of different division types, obtain the center points of each dominant structural group based on the attitude information. The angle between the position of the center point in the plane coordinate system and the y-axis is the dip direction, and the radius length is the dip angle. Extract the attitude information of the center point to obtain the structural plane parameters. Identify the dangerous rock mass based on the structural plane parameters and the collapse stability judgment table.

[0114] Among them, the specific steps for identifying dangerous rock masses include: obtaining the spatial distribution characteristics of the dominant structural planes, obtaining the first structural planes that intersect with each rock mass based on the spatial distribution characteristics, and obtaining the number of structural planes of the first structural planes; obtaining the attitude relationship between the first structural planes and the rock masses based on the structural plane parameters; obtaining the rock mass collapse information based on the number of structural planes, the attitude relationship, and the collapse stability judgment table; and identifying dangerous rock masses based on the rock mass collapse information.

[0115] In this embodiment, the collapse stability judgment table can be as shown in Table 1:

[0116] Table 1 Collapse Stability Judgment Table

[0117]

[0118]

[0119] Among them, the specific steps for identifying dangerous rock masses based on the rock mass collapse information include: obtaining the collapsed rock masses based on the rock collapse information, extracting the outcrop points of the collapsed rock masses; obtaining the second structural planes of the collapsed rock masses, and extracting the structural plane intersection lines of the second structural planes; obtaining the spatial coordinates and collapse numbers based on the outcrop points and the structural plane intersection lines, and marking the dangerous rock masses based on the spatial coordinates and the collapse numbers.

[0120] In this embodiment, manual review can also be combined. For example, for the automatically identified collapse characteristics and spatial position information, manual review is combined with point cloud data and three-dimensional image models to correct the incorrect judgments, and finally determine the regional collapse quantity, spatial position, and other information characteristics.

[0121] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be interpreted to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0122] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for identifying dangerous rock masses in high and steep mountainous areas, characterized in that: The method comprises: Acquire first image data of the target area based on a first acquisition device, and construct a coarseness model based on the first image data; Obtain flight route parameters based on the first image data, collect second image data of the target area based on the flight route parameters and a second acquisition device, and construct a three-dimensional image model and a point cloud model based on the second image data and the roughness model; Obtaining a dominant structural surface based on the three-dimensional image model and the point cloud model, and obtaining structural surface parameters based on the dominant structural surface; Identify dangerous rock masses based on the structural surface parameters and the collapse stability judgment table; The specific steps to obtain flight route parameters include: Acquire first device parameters of the first acquisition device, and obtain first parameters based on the first device parameters, wherein the first parameters include ground coverage and ground resolution, and the first device parameters include a focal length of the first acquisition device, a width and height of a sensor of the first acquisition device, and a flying height and pixel size of the first acquisition device; Obtaining a shooting distance between two adjacent images based on the first image data, and obtaining a second parameter based on the shooting distance and the ground coverage, wherein the second parameter includes a heading overlap, a lateral overlap, and a route distance; Acquire the area size of the target area, and obtain a third parameter based on the area size, the shooting interval and the route interval, wherein the third parameter includes the number of routes and the number of shooting points of each route; Obtaining the flight route parameter based on the first parameter, the second parameter and the third parameter; The specific steps to obtain the advantageous structural surface include: Acquire the point clouds of the three-dimensional image model and the point cloud model, search for the nearest neighbor points of the point cloud to obtain a nearest neighbor point set; obtain a covariance matrix based on the nearest neighbor point set, and obtain eigenvalues ​​based on the covariance matrix; Obtaining a deviation parameter based on the eigenvalue, and obtaining a first point cloud set based on the deviation parameter and the maximum deviation; obtaining a point normal vector of the first point cloud set based on the eigenvalue; and unifying the direction of the point normal vector based on a preset direction to obtain a coplanar point cloud set; Obtaining the dip angle and inclination based on the coplanar point cloud set and the plane equation; obtaining the occurrence of the structural surface based on the dip angle, the inclination and the point normal vector; Performing point cloud density calculation on the coplanar point cloud set to obtain kernel density, performing density division on the first point cloud set based on the kernel density to obtain a plurality of density clusters, obtaining a maximum kernel density of each density cluster based on the kernel density, obtaining a density peak point based on the maximum kernel density, and performing cluster analysis on each density cluster based on the density peak point to obtain the dominant structural surface; The specific steps of performing cluster analysis on the density cluster based on the density peak point include: S1. arbitrarily select an area to obtain a seed area, and obtain an adjacent area adjacent to the seed area; S2, respectively calculating the grayscale means of the seed region and the adjacent region, and obtaining similarity based on the grayscale means; S3, judging whether the similarity is less than a preset value, if so, obtaining the dominant structural surface based on the seed region and the adjacent region; if not, merging the seed region and the adjacent region to obtain a first region, returning to S2, and updating the seed region to the first region; The specific steps to identify dangerous rock masses include: Acquire the spatial distribution characteristics of the dominant structural surface, obtain the first structural surface that has an intersecting relationship with each rock mass based on the spatial distribution characteristics, and obtain the number of structural surfaces of the first structural surface; Obtaining the occurrence relationship between the first structural surface and the rock mass based on the structural surface parameters; Obtain rock collapse information based on the number of structural planes, the occurrence relationship and the collapse stability judgment table; identify dangerous rock mass based on the rock collapse information; The specific steps of identifying dangerous rock mass based on the rock collapse information include: Obtaining collapsed rock mass based on the rock mass collapse information, and extracting an outcropping point of the collapsed rock mass; Acquire a second structural surface of the collapsed rock mass, and extract a structural surface intersection line of the second structural surface; The spatial coordinates and the collapse number are obtained based on the outcropping point and the intersection line of the structural surface, and the dangerous rock mass is marked based on the spatial coordinates and the collapse number.

2. A method for identifying dangerous rock masses in high and steep mountainous areas according to claim 1, characterized in that: The specific steps of building a 3D image model and a point cloud model include: Acquire second device parameters of the second acquisition device, and obtain internal orientation parameters based on the second device parameters, wherein the second device parameters include focal length and principal point coordinates; Matching the second image data based on preset feature points to obtain corresponding points, and obtaining external orientation parameters based on the corresponding points, wherein the external orientation parameters include a rotation matrix and a translation vector; Acquire fourth parameters of the second image data, and obtain three-dimensional space coordinates based on the corresponding points and the fourth parameters, wherein the fourth parameters include a projection matrix and a scale factor; Constructing a projection equation based on the internal orientation parameter, the external orientation parameter and the three-dimensional space coordinates; Constructing a triangulated network based on the roughness model and the projection equation; An optimal texture image is obtained based on the second image data, and texture mapping is performed on each triangular face in the triangulated network based on the optimal texture image to obtain the three-dimensional image model and the point cloud model.

3. A method for identifying dangerous rock masses in high and steep mountainous areas according to claim 2, characterized in that: Before constructing the triangulated network, the method further includes: optimizing the projection equation based on minimizing the reprojection error, and constructing the triangulated network based on the roughness model and the optimized projection equation.

4. The method for identifying dangerous rock masses in high and steep mountainous areas according to claim 2 is characterized in that: The specific steps of constructing a triangulated network include: Triangulating the roughness model to obtain a triangular mesh; dividing the triangular mesh into a plurality of polygons to obtain Thiessen polygons; A normal vector is obtained based on preset triangle vertices; and based on the normal vector and the projection equation, surface reconstruction and surface fitting are performed on the Thiessen polygon to obtain the triangulated network.

5. The method for identifying dangerous rock masses in high and steep mountainous areas according to claim 1, characterized in that: The specific steps of obtaining the structural surface parameters include: Based on the threshold and the occurrence, the dominant structural surface is sequentially fused, corrected and occurrence information is extracted to obtain the structural surface parameters.

Images