A method and system for constructing dangerous rock mass identification and extraction rules taking into account multi-dimensional characteristics

Abstract

The present invention discloses a method and system for constructing dangerous rock identification and extraction rules taking into account multi-dimensional features, which belongs to the field of landslide-type dangerous rock identification and early warning, wherein the method for constructing dangerous rock identification and extraction rules includes the following steps: based on point cloud data, an improved progressively encrypted triangulated network filtering algorithm and a moving curve fitting filtering algorithm are proposed to filter vegetation, and ground point clouds are extracted in combination with remote sensing images; based on the development characteristics of dangerous rock bodies, dangerous rock identification and interpretation marks taking into account the development characteristics are established; based on the mapping characteristics of identification indicators, single feature calculations of different scales are performed to establish multi-scale identification and extraction rules for dangerous rock bodies; and a dangerous rock identification and extraction rule construction system taking into account multi-dimensional features is designed and developed. The present invention establishes a dangerous rock interpretation mark and a method for selecting characteristic parameters, and constructs dangerous rock identification and extraction rules taking into account multi-dimensional features, which has certain scientific value and practical significance for quantitative assessment of road-related dangerous rock disaster risks and disaster prevention and mitigation management.

Description

Technical Field

[0001] The invention relates to the technical field of landslide-type dangerous rock mass identification and early warning, and in particular to a dangerous rock mass identification and extraction rule construction method and system taking multi-dimensional features into consideration. Background Art

[0002] Dangerous rock disasters refer to geological disasters in which rocks are broken and disintegrated due to weathering, erosion, earthquakes, gravity and other factors due to natural factors or human activities, and may break away from the parent rock and roll or fall downward, posing a threat to personnel and facilities below. Such disasters are common in mountainous areas, river valleys and areas with frequent human engineering activities.

[0003] At present, the situation of dangerous rock disaster prevention and control in my country is still severe, and the accurate identification and rapid extraction of dangerous rocks are the important technical basis for the monitoring and prevention of dangerous rock disasters. Dangerous rock masses have diverse causes, complex structures, and are extremely hidden and sudden. It is difficult to identify and extract them. There is a lack of quantitative signs that can reflect the main controlling factors of dangerous rock masses, and the refined expression of typical development characteristics such as free surfaces and concave cavities of dangerous rock masses is still one of the difficulties in the identification and extraction of dangerous rocks. Therefore, how to select the main controlling factors of dangerous rock masses and construct dangerous rock mass interpretation signs that take into account multi-dimensional characteristics is one of the key issues in the identification and extraction of dangerous rocks, and how to use three-dimensional laser scanning data to realize the rapid identification of structures such as free surfaces and concave cavities of dangerous rock masses and accurately extract the point cloud of suspected dangerous rock masses is another key issue. Summary of the invention

[0004] In order to solve the problems in the above-mentioned background technology, the present invention provides a method and system for constructing dangerous rock body identification and extraction rules taking into account multi-dimensional features. In view of a series of problems such as diverse dangerous rock types, large temporal and spatial differences in development characteristics, complex structure, and difficulty in monitoring and early warning, the present invention establishes dangerous rock body interpretation signs and characteristic parameter selection, and constructs dangerous rock body identification and extraction rules taking into account multi-dimensional features. This provides a new method for dangerous rock identification, diagnosis and dynamic monitoring, and has certain scientific value and practical significance for quantitative assessment of dangerous rock disaster risks and disaster prevention and mitigation management in road areas.

[0005] To achieve the above object, the present invention provides a method for constructing dangerous rock mass identification and extraction rules taking into account multi-dimensional features, comprising the following steps:

[0006] Step S1: Based on the three-dimensional laser point cloud data, an improved progressive encryption triangulation filtering algorithm and a moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution remote sensing images, the filtered point cloud classification results are checked and corrected by a manual auxiliary verification method, and the ground point cloud is extracted by point cloud attributes;

[0007] Step S2: Based on the development characteristics of dangerous rock masses, establish dangerous rock mass identification and interpretation marks that take into account the development characteristics;

[0008] Step S3: Based on the mapping characteristics of the identification index, the calculation and optimization of single features at different scales are performed respectively, and the selection of feature scales of dangerous rock mass point cloud and its threshold are proposed to establish a multi-scale identification and extraction rule for dangerous rock mass;

[0009] Step S4: Design and develop a dangerous rock mass identification and extraction rule construction system that takes into account multi-dimensional features.

[0010] Preferably, step S1 comprises the following steps:

[0011] S101, according to the location of the dangerous rock body, the scope of the survey area, the performance of the 3D laser scanner, the lens scanning width and the surrounding environment, determine the distance to the dangerous rock body and select the point with the widest scanning angle to mark as a fixed observation point;

[0012] S102, setting up a three-dimensional laser scanner at a fixed observation point, setting scanning parameters and performing station scanning;

[0013] S103, based on S101-S102, original point cloud data is obtained, and then measured coordinates are uniformly converted, registered, spliced ​​and cropped in sequence;

[0014] S104, using a statistical filtering algorithm to process outlier noise, and proposing an improved progressive encryption triangulation filtering algorithm and a moving curve fitting filtering algorithm to filter vegetation, combined with high-resolution remote sensing images, to obtain a denoised point cloud;

[0015] S105, using a manual assisted verification method to calibrate and correct the filtered point cloud classification results, and extracting the ground point cloud through point cloud attributes.

[0016] Preferably, in step S104, the improved progressively encrypted triangulated network filtering algorithm and the moving curve fitting filtering algorithm include the following steps:

[0017] Ground seed point extraction: Thin plate spline interpolation TPS and cloth simulation CSF methods are used to obtain ground points respectively. The obtained ground seed points are fused and further selected according to the terrain, point density and data quality conditions to select the lowest point that is suitable as the seed point set;

[0018] The local neighborhood method optimizes and constructs the initial triangulated network that is closest to the terrain features. Based on the regional growth concept, the progressive triangulated network is encrypted step by step in the subdivided area of ​​the minimum slope unit.

[0019] A Gaussian mixture model that takes terrain classification into account is used to complete reference surface fitting at different levels;

[0020] The surfaces generated by the progressively encrypted triangulated network filtering and the moving curve fitting are fused, compared and optimized, and the parameters are gradually fine-tuned according to the conditions of different point cloud data, and the threshold is adaptively and dynamically optimized;

[0021] Manual assisted verification is used to check whether there are non-ground point clouds.

[0022] Preferably, in step S2, the development characteristics of the dangerous rock mass include steep slope, broken rock mass, developed cracks, large free surface and deep concave cavity.

[0023] Preferably, step S2 comprises the following steps:

[0024] S201. Define the critical point based on the slope value range of the given study area and field survey results;

[0025] S202, judging the previous and current degree of fragmentation and unloading cracks of the rock mass by taking photos in combination with field investigation;

[0026] S203, observing and obtaining the depth of the cavity from different angles on photos taken during field investigation;

[0027] S204. Obtain the size of the free surface based on the field investigation of the rock mass, find the turning point based on historical data, and construct judgment conditions.

[0028] Preferably, in S201, the slope is used to describe the possibility of the dangerous rock mass;

[0029] In S202, the degree of fragmentation and unloading cracks characterize the stability of dangerous rock mass. The greater the degree of rock fragmentation and the opening of unloading cracks, the worse its stability is; the smaller the degree of rock fragmentation and the opening of unloading cracks, the stronger its stability is.

[0030] In S203, the concave cavity easily provides space conditions for the rock mass. The greater the depth of the concave cavity, the worse the stability of the rock mass.

[0031] In S204, the size of the free surface area determines the stability of the dangerous rock mass. The larger the free surface area, the worse the stability of the dangerous rock mass.

[0032] Preferably, step S3 comprises the following steps:

[0033] S301, calculating the slope based on the high-precision digital elevation model DEM;

[0034] S302, computing an inclination angle D based on ground point computing;

[0035] S303, according to the surface characteristics and linear characteristics of the dangerous rock mass fragmentation and the unloading cracks, by calculating the covariance matrix eigenvalues ​​of the local neighborhood of the dangerous rock mass point cloud, the combined information of the point cloud covariance eigenvalues ​​is obtained;

[0036] S304, respectively obtain three eigenvalues ​​and corresponding eigenvectors through S303, and combine the two to calculate the planarity feature of the point cloud;

[0037] S305, calculating the total variance according to the shadow characteristics of the sharp part of the concave cavity terrain at the rear edge of the dangerous rock;

[0038] S306. Based on the spatial state of the dangerous rock mass, which is characterized by the free surface, the relationship between the fitting plane and the horizontal plane at a certain point is described by verticality;

[0039] S307, based on the three-dimensional spatial features of the point cloud, display the effects at small and large scales, formulate judgment conditions, and comprehensively select the best threshold value based on the judgment conditions and field investigation to achieve the best results;

[0040] S308. Based on S301~S307, a multi-scale identification and extraction rule for dangerous rock masses is constructed and applied to the identification and early warning of landslide-type dangerous rock masses in high mountain canyon areas.

[0041] Preferably, in S301, the slope calculation formula is as follows:

[0042] ;

[0043] in, Indicates the slope value measured in degrees. , Represent the increments in the horizontal and vertical directions respectively, Yes is the value obtained by truncating the calculation result;

[0044] In S302, the inclination angle calculation formula is as follows:

[0045] ;

[0046] In the formula, is the projection length of the external normal vector of the structural surface on the axis, ;

[0047] In S303, it is assumed that the point to be measured in the point cloud data is ,by is the center of the sphere and the radius is The neighborhood point set in the ball is , then the point to be tested The covariance matrix of The calculation formula is:

[0048] ;

[0049] ;

[0050] ;

[0051] in, For this neighborhood The coordinates of the points, For the The number of points in the neighborhood of a point, is the centroid of the neighborhood;

[0052] In S304, the three eigenvalues ​​are , the eigenvectors corresponding to the three eigenvalues ​​are , point cloud plane characteristics The formula is:

[0053] ;

[0054] In S305, the total variance The formula is:

[0055] ;

[0056] In S306, verticality The formula is:

[0057] ;

[0058] S307, the judgment condition is: when calculating the local features of the point cloud, the size of the spherical neighborhood search radius r is considered. If the search radius r is small, the number of points involved in the calculation will be small. If the search radius r is large, the data calculation time will be too long.

[0059] The present invention also provides a system for constructing a dangerous rock mass identification and extraction rule method taking into account multi-dimensional features, comprising:

[0060] Data processing and algorithm improvement module: Based on the original 3D laser point cloud data, an improved progressive encryption triangulation filtering algorithm and moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution images, a human-computer interactive method is used to check and correct the denoised point cloud to obtain the real ground point cloud;

[0061] Interpretation sign construction module taking into account the development characteristics of dangerous rock masses: Based on the development characteristics of dangerous rock masses, interpretation signs for identifying dangerous rock masses taking into account the development characteristics are established from different perspectives;

[0062] The module for comparing and selecting the characteristic scales of dangerous rock mass point cloud and determining its thresholds and establishing the rules for multi-scale identification and extraction of dangerous rock mass is as follows: the optimal thresholds for the inclination, planarity, total variance and verticality characteristics of dangerous rock mass are calculated and selected from different scales, and based on this, the rules for multi-scale identification and extraction of dangerous rock mass are established.

[0063] Therefore, the present invention adopts the above-mentioned dangerous rock body identification and extraction rule construction method and system taking into account multi-dimensional features, which has the following beneficial effects:

[0064] (1) In view of the diverse types of dangerous rocks, large spatiotemporal differences in development characteristics, and complex structures, the present invention has established dangerous rock mass interpretation signs that take into account multi-dimensional features such as slope, fragmentation, cracks, concave cavities, and free surfaces based on three-dimensional laser point cloud data and image data. This has improved and ensured the comprehensiveness and accuracy of the dangerous rock mass interpretation sign samples, and provided a scientific basis for the main controlling factors and quantitative indicators for dangerous rock mass identification.

[0065] (2) Through the quantitative calculation of features at various scales and the optimization of feature parameters, a dangerous rock mass identification and extraction rule that takes into account multi-dimensional features is constructed. Compared with a single feature parameter, a more accurate new rule and method is provided for the identification, diagnosis and dynamic monitoring of dangerous rocks. It has a positive practical effect on the accuracy and efficiency of accurate identification of dangerous rock masses on roads in alpine canyon areas and quantitative assessment of disaster risks.

[0066] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 is a flowchart of steps of an embodiment of the present invention;

[0068] Figure 2 A technical roadmap for an embodiment of the present invention;

[0069] Figure 3 A schematic diagram of the dangerous rock mass identification characteristics according to an embodiment of the present invention;

[0070] Figure 4 This is a relationship diagram between the slope and inclination of the rock mass terrain in an embodiment of the present invention. Figure 4 (a) is the rock mass topography slope map, Figure 4 (b) is the rock mass topography dip map;

[0071] Figure 5 For the different embodiments of the present invention r Rock plane characteristic result diagram under value;

[0072] Figure 6 For the different embodiments of the present invention r The rock full variance characteristic result diagram under the value;

[0073] Figure 7 For the different embodiments of the present invention r Rock verticality characteristic result diagram under the value;

[0074] Figure 8 A schematic diagram of the system structure of an embodiment of the present invention;

[0075] Fig. 9 Schematic diagram of the inclination concept of an embodiment of the present invention. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical scheme and advantages disclosed in the embodiments of the present invention clearer, the embodiments of the present invention are further described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not intended to limit the embodiments of the present invention. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions.

[0077] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or inherent to these processes, methods, products or devices.

[0078] Like reference numerals and letters denote similar items in the following drawings, and thus, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0079] In the description of the present invention, it should be noted that the terms "upper", "lower", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, or are directions or positional relationships in which the product of the invention is usually placed when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as a limitation on the present invention.

[0080] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "setting", "installation" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0081] Example

[0082] like Figure 1 , Figure 2 As shown, the method for constructing dangerous rock mass identification and extraction rules taking into account multi-dimensional features of the present invention comprises the following steps:

[0083] Step S1: Based on the three-dimensional laser point cloud data, an improved progressive encryption triangulation filtering algorithm and moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution remote sensing images, the filtered point cloud classification results are checked and corrected by manual assisted verification, and the ground point cloud is extracted through point cloud attributes.

[0084] Step S1 includes the following steps:

[0085] S101, according to the specific location of the dangerous rock body, the scope of the survey area, the performance of the 3D laser scanner, the lens scanning width and the surrounding environment, select the point with the widest scanning angle at a certain distance from the dangerous rock body and mark it as a fixed observation point;

[0086] S102, setting up a three-dimensional laser scanner at a fixed observation point, setting scanning parameters and performing station scanning;

[0087] S103, based on S101-S102, original point cloud data is obtained, and then measured coordinates are uniformly converted, registered, spliced ​​and cropped in sequence;

[0088] S104, using a statistical filtering algorithm to process outlier noise, and proposing an improved progressive encryption triangulation filtering algorithm and a moving curve fitting filtering algorithm to filter vegetation, combined with high-resolution remote sensing images, to obtain a denoised point cloud;

[0089] Statistical filtering algorithms are used to remove obvious outliers. The characteristic of outliers is that they are sparsely distributed in space. Laser scanning usually generates point cloud data sets with different point densities, and measurement errors can also lead to sparse outliers / outliers. Taking into account the characteristics of outliers, it can be defined that if the point cloud is less than a certain density, the point cloud is invalid. The specific steps are to calculate the average distance from each point to its nearest points. By assuming that the distance result distribution of its points in the point cloud is a Gaussian distribution with a mean and standard deviation, points with an average distance outside the standard range can be defined as outliers and removed from the data according to the given mean and variance.

[0090] The specific steps of the improved progressive encryption triangulation filtering algorithm and moving curve fitting filtering algorithm are as follows:

[0091] 1) Ground seed point extraction: Thin plate spline interpolation (TPS) and cloth simulation (CSF) are used to obtain ground points respectively, and all the obtained ground seed points are integrated to further select the appropriate lowest point as the seed point set according to the terrain, point density and data quality conditions;

[0092] Thin plate spline interpolation function:

[0093] Thin Plate Spline (TPS) is a commonly used two-dimensional interpolation method, and its formula is as follows:

[0094] ;

[0095] in, are the coordinates of the difference point. are the coordinates of the known points, are the coordinates of the next point, is the total number of points, , is the weight, is the basis function, usually chosen as a two-dimensional radial basis function, such as a Gaussian function:

[0096] ;

[0097] in is the distance from the control point to the point to be interpolated. This formula determines the interpolation function by minimizing the bending energy, solving the problem of excessive fluctuation of the interpolation function. In three-dimensional interpolation, it is also necessary to add Item, and satisfy the equation after( represents a number), decompose the formula, construct a matrix and take its inverse, and multiply the matrix on the left Coordinate value.

[0098] Cloth simulation algorithm:

[0099] Cloth Simulation Filtering (CSF) is a method used for point cloud data processing. Its basic principle is to invert the point cloud and cover the surface of the point cloud with a layer of cloth. The cloth will fall due to gravity. If the cloth is soft enough, it can completely adhere to the inverted point cloud surface and finally form a digital surface model. If the cloth is hard, a digital ground model will be formed.

[0100] The cloth simulated by the computer is a grid composed of a large number of discrete nodes. The nodes in the grid are interconnected and connected by connecting lines. At the same time, it is assumed that each grid point is a particle with constant mass and no size. Therefore, this grid method is also called a particle spring model. To simulate the shape of the cloth at a certain moment in free fall, it is necessary to calculate the spatial position of each cloth point. It is assumed that each cloth node moves only in the vertical direction and is only subject to two forces: gravity and the interaction force of the neighboring nodes. When only the effect of gravity is considered, it can be known from Newton's second law that the relationship between the spatial position of the cloth point and the force it is subjected to can be determined by the following formula:

[0101] ;

[0102] in, m Indicates the quality of the cloth point, Gis the gravitational acceleration constant, X(t) represents the grid node position at a certain moment, △t Indicates the time step. When the time step and the initial position of the grid node are known, the current position of the grid node can be calculated.

[0103] In order to constrain the reversal of particles in the blank area of ​​the reversal surface, the displacement caused by internal factors of the particles must also be considered. Randomly select two adjacent particles. If both particles are movable, move them the same distance in opposite directions; if one is immovable, move the other; if the two particles have the same height, do not move. The displacement can be calculated by the following formula:

[0104] ;

[0105] in, d is the displacement of the particle; when the particle is movable, b When equal to 1, it cannot be moved b is 0; for The neighboring particles; n is the unit vector normalized to the vertical direction. .

[0106] The detailed steps of the cloth simulation filtering algorithm are as follows:

[0107] ① Flip the original point cloud upside down and transform the point cloud space coordinates.

[0108] ② Set the grid resolution and virtually grid the original point cloud. Project all laser points and particles on the grid on the same horizontal plane, find the nearest point of each particle, and record its elevation before projection.

[0109] ③ For each movable grid particle, calculate its displacement caused by gravity, and compare the position with the elevation of its corresponding laser foot point. If the elevation of the cloth grid node is less than or equal to the elevation of the laser foot point, move the node to the position of the laser foot point and mark it as an immovable point.

[0110] ④ Calculate the displacement that each cloth grid node needs to move after being acted upon by the forces from the adjacent points.

[0111] ⑤ Repeat steps 3 and 4. When the elevation changes of each node begin to become small enough or the number of iterations reaches the maximum, the cloth simulation process can be ended.

[0112] ⑥ Distinguish between ground points and non-ground points. If the distance between the LiDAR point and the simulated particle is less than the preset threshold, it is considered to be a ground point, otherwise it is considered to be a non-ground point.

[0113] 2) The local neighborhood method is used to optimize and construct the initial triangulated network that is closest to the terrain features. Based on the regional growth concept, the progressive triangulated network is gradually encrypted in the subdivided area of ​​the minimum slope unit to improve the filtering accuracy at the ridges and valleys.

[0114] 3) Adopting the Gaussian mixture model taking into account terrain classification, the reference surfaces at different levels are accurately fitted;

[0115] Gaussian Mixture Model

[0116] Gaussian Mixture Model (GMM) is a model used to describe data distribution. It represents the probability distribution of a random variable by weighting multiple Gaussian distributions. In terrain classification, Gaussian mixture models can be used to deal with the complexity and diversity of terrain data. According to the search results, the formula of Gaussian mixture model can be listed in detail as follows:

[0117] ;

[0118] in, x is the observed value, μ is the mean, σ is the standard deviation.

[0119] ;

[0120] in, x yes d dimensional sample observation vector, is the mixing weight, and , yes d dimensional Gaussian density function, are the corresponding parameters; are the Gaussian mixture model parameters.

[0121] Gaussian mixture model parameter estimation is based on the observed samples of the mixture distribution parameters The process of making an estimate. This is usually done by maximizing expectation ( EM ) algorithm, which consists of two steps: E step (expected step) and M step (maximization step). E In the step, the probability (responsibility value) of each data point belonging to each Gaussian component is calculated. M In the next step, the parameters of the Gaussian components are updated according to these responsibility values.

[0122] 4) The surfaces generated by the progressively encrypted triangulated network filtering and the moving curve fitting are fused, compared and optimized, and the parameters are gradually fine-tuned according to the specific conditions of different point cloud data, and the threshold is adaptively and dynamically optimized to ensure more accurate identification and filtering of vegetation points. This combined method not only improves the filtering accuracy, but also enhances the adaptability and robustness of the algorithm.

[0123] S105, the filtered point cloud classification results are checked and corrected by the manual assisted verification method, and the ground point cloud is extracted by the point cloud attributes. The manual assisted verification mainly checks whether there are non-ground point clouds, such as vegetation, towers, power lines, road projects and their ancillary facilities, houses, etc. If they exist, they need to be removed, and vice versa.

[0124] Step S2: Based on the development characteristics of dangerous rock mass, including steep slope, broken rock mass, developed cracks, large free surface and deep cavity, such as Figure 3 As shown, the identification and interpretation signs of dangerous rock bodies taking into account the development characteristics are established.

[0125] Step S2 includes the following steps:

[0126] S201, slope is used to describe the possibility of dangerous rock mass danger. The critical point is defined according to the slope value range of the given study area and the field investigation situation;

[0127] S202. Fragmentation and unloading cracks characterize the stability of dangerous rock mass. If the rock mass fragmentation and unloading crack opening are larger, the stability is worse, and vice versa. It is necessary to combine field survey and take photos to judge the previous and current fragmentation and unloading crack conditions of the rock mass.

[0128] S203. The cavity can easily provide space conditions for the rock mass. That is, the greater the depth of the cavity, the worse the stability of the rock mass. The depth of the cavity can be observed from different angles in the field survey photos.

[0129] S204. The size of the free surface area determines the stability of the dangerous rock mass. That is, the larger the free surface area, the more unstable the dangerous rock mass. The size of the free surface area is obtained based on the field investigation of the rock mass, combined with historical data, to find the turning point and construct the judgment conditions.

[0130] Step S3: Based on the mapping characteristics of the identification index, the calculation and optimization of single features at different scales are performed respectively, the selection of feature scales of dangerous rock mass point cloud and its threshold are proposed, and the extraction rules for multi-scale identification of dangerous rock mass are established.

[0131] Step S3 includes the following steps:

[0132] S301. Calculate the slope based on a high-precision digital elevation model (DEM). The formula is as follows:

[0133] ;

[0134] in, Indicates the slope value measured in degrees. , Represent the increments in the horizontal and vertical directions respectively, Yes is the value obtained by truncating the calculation result;

[0135] S302, Inclination angle based on ground point computing , similar to the slope method and principle, such as Figure 4 As shown;

[0136] like Fig. 9 As shown in the figure, the structural surface attitude is the information that directly indicates the spatial position of the structural surface, which is mainly composed of three elements: strike, dip and dip. The strike usually refers to the extension direction of the intersection line of the discontinuity surface and the horizontal plane; the dip is the line on the structural surface that is perpendicular to the strike and points downward along the discontinuity surface; the dip is the angle between the dip and the structural surface.

[0137] The formula for calculating the inclination angle (°) is as follows:

[0138] ;

[0139] In the formula, is the projection length of the external normal vector of the structural surface on the axis, .

[0140] S303, according to the surface characteristics and linear characteristics of the dangerous rock mass fragmentation and the unloading cracks, by calculating the covariance matrix eigenvalues ​​of the local neighborhood of the dangerous rock mass point cloud, the combined information of the point cloud covariance eigenvalues ​​is obtained;

[0141] The calculation principle is as follows: Assume that the point to be measured in the point cloud data is ,by is the center of the sphere and the radius is The neighborhood point set in the ball is , then the point to be tested The covariance matrix of The calculation formula is:

[0142] ;

[0143] ;

[0144] ;

[0145] in, For this neighborhood The coordinates of the points, For the The number of points in the neighborhood of a point, is the centroid of the neighborhood.

[0146] S304, such as Figure 5 As shown, three eigenvalues ​​are obtained respectively through S303 ( ) and the corresponding eigenvector ( ), and combine the two to calculate the point cloud plane characteristics , the formula is as follows:

[0147] ;

[0148] S305, such as Figure 6 As shown in the figure, according to the sharp shadow characteristics of the concave cavity terrain at the rear edge of the dangerous rock, the total variance is calculated , the formula is as follows:

[0149] ;

[0150] S306, such as Figure 7 As shown in the figure, according to the spatial state of the dangerous rock mass, it is characterized by the free surface, and the verticality ( ) describes the relationship between the fitted plane and the horizontal plane at a certain point, verticality The formula is:

[0151] ;

[0152] S307. Based on the three-dimensional spatial features of point clouds, the effects are demonstrated at small and large scales. Based on this, when calculating the local features of point clouds, the size of the spherical neighborhood search radius r is considered. If the search radius r is too small, the number of points involved in the operation will be small. Otherwise, the data calculation time will be too long and the meaning described by the point cloud features will be difficult to show. Based on the above judgment conditions and field investigations, the best threshold is comprehensively selected to achieve the best results.

[0153] S308. Based on S301~S307, a multi-scale identification and extraction rule for dangerous rock masses is constructed and applied to the identification and early warning of landslide-type dangerous rock masses in high mountain canyon areas.

[0154] Step S4: Design and develop a dangerous rock mass identification and extraction rule construction system that takes into account multi-dimensional features.

[0155] like Figure 8 As shown, the system of the dangerous rock body identification and extraction rule construction method taking into account multi-dimensional features of the present invention includes:

[0156] Data processing and algorithm improvement module: Based on the original 3D laser point cloud data, an improved progressive encryption triangulation filtering algorithm and moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution images, a human-computer interactive method is used to check and correct the denoised point cloud to obtain the real ground point cloud;

[0157] Interpretation sign construction module taking into account the development characteristics of dangerous rock masses: Based on the development characteristics of dangerous rock masses, interpretation signs for identifying dangerous rock masses taking into account the development characteristics are established from different perspectives;

[0158] The module for comparing and selecting the characteristic scales of dangerous rock mass point cloud and determining its thresholds and establishing the rules for multi-scale identification and extraction of dangerous rock mass is as follows: the optimal thresholds for the inclination, planarity, total variance and verticality characteristics of dangerous rock mass are calculated and selected from different scales, and based on this, the rules for multi-scale identification and extraction of dangerous rock mass are established.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for constructing dangerous rock mass identification and extraction rules taking into account multi-dimensional features, characterized in that: The following steps are involved: Step S1: Based on the three-dimensional laser point cloud data, an improved progressive encryption triangulation filtering algorithm and a moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution remote sensing images, the filtered point cloud classification results are checked and corrected by a manual auxiliary verification method, and the ground point cloud is extracted by point cloud attributes; Step S2: Based on the development characteristics of dangerous rock masses, establish dangerous rock mass identification and interpretation marks that take into account the development characteristics; Step S3: Based on the mapping characteristics of the identification index, the calculation and optimization of single features at different scales are performed respectively, and the selection of feature scales of dangerous rock mass point cloud and its threshold are proposed to establish a multi-scale identification and extraction rule for dangerous rock mass; Step S4: designing and developing a dangerous rock mass identification and extraction rule construction system taking into account multi-dimensional features; Step S1 includes the following steps: S101, according to the location of the dangerous rock body, the scope of the survey area, the performance of the 3D laser scanner, the lens scanning width and the surrounding environment, determine the distance to the dangerous rock body and select the point with the widest scanning angle to mark as a fixed observation point; S102, setting up a three-dimensional laser scanner at a fixed observation point, setting scanning parameters and performing station scanning; S103, based on S101-S102, original point cloud data is obtained, and then measured coordinates are uniformly converted, registered, spliced ​​and cropped in sequence; S104, using a statistical filtering algorithm to process outlier noise, and proposing an improved progressive encryption triangulation filtering algorithm and a moving curve fitting filtering algorithm to filter vegetation, combined with high-resolution remote sensing images, to obtain a denoised point cloud; S105, using a manual assisted verification method to calibrate and correct the filtered point cloud classification result, and extracting the ground point cloud according to the point cloud attributes; In step S104, the improved progressively encrypted triangulated network filtering algorithm and the moving curve fitting filtering algorithm include the following steps: Ground seed point extraction: Thin plate spline interpolation TPS and cloth simulation CSF methods are used to obtain ground points respectively. The obtained ground seed points are fused and further selected according to the terrain, point density and data quality conditions to select the lowest point that is suitable as the seed point set; The local neighborhood method optimizes and constructs the initial triangulated network that is closest to the terrain features. Based on the regional growth concept, the progressive triangulated network is encrypted step by step in the subdivided area of ​​the minimum slope unit. A Gaussian mixture model that takes terrain classification into account is used to complete reference surface fitting at different levels; The surfaces generated by the progressively encrypted triangulated network filtering and the moving curve fitting are fused, compared and optimized, and the parameters are gradually fine-tuned according to the conditions of different point cloud data, and the threshold is adaptively and dynamically optimized; Manual assisted verification is used to check whether there are non-ground point clouds; In step S2, the development characteristics of dangerous rock mass include steep slope, broken rock mass, developed cracks, large free surface and deep cavity; Step S2 includes the following steps: S201. Define the critical point based on the slope value range of the given study area and field survey results; S202, judging the previous and current fragmentation and unloading crack conditions of the rock mass by taking photos in combination with field investigation; S203, observing and obtaining the depth of the cavity from different angles on photos taken during field investigation; S204, obtaining the size of the free surface according to the field investigation of the rock mass, finding the turning point and constructing the judgment condition in combination with the historical data; In S201, slope is used to describe the potential magnitude of dangerous rock mass hazards; In S202, the degree of fragmentation and unloading cracks characterize the stability of dangerous rock mass. The greater the degree of rock fragmentation and the opening of unloading cracks, the worse its stability is; the smaller the degree of rock fragmentation and the opening of unloading cracks, the stronger its stability is. In S203, the concave cavity easily provides space conditions for the rock mass. The greater the depth of the concave cavity, the worse the stability of the rock mass. In S204, the size of the free surface area determines the stability of the dangerous rock mass. The larger the free surface area, the worse the stability of the dangerous rock mass. Step S3 includes the following steps: S301, calculating the slope based on the high-precision digital elevation model DEM; S302, computing an inclination angle D based on ground point computing; S303, according to the surface characteristics and linear characteristics of the dangerous rock mass fragmentation and the unloading cracks, by calculating the covariance matrix eigenvalues ​​of the local neighborhood of the dangerous rock mass point cloud, the combined information of the point cloud covariance eigenvalues ​​is obtained; S304, respectively obtaining three eigenvalues ​​and corresponding eigenvectors through S303, and combining the two to calculate the planarity feature of the point cloud; S305, calculating the total variance according to the shadow characteristics of the sharp part of the concave cavity terrain at the rear edge of the dangerous rock; S306. Based on the spatial state of the dangerous rock mass, which is characterized by the free surface, the relationship between the fitting plane and the horizontal plane at a certain point is described by verticality; S307, based on the three-dimensional spatial features of the point cloud, display the effects at small and large scales, formulate judgment conditions, and comprehensively select the best threshold value based on the judgment conditions and field investigation to achieve the best results; S308. Based on S301~S307, a multi-scale identification and extraction rule for dangerous rock masses is constructed and applied to the identification and early warning of landslide-type dangerous rock masses in high mountain canyon areas.

2. The method for constructing dangerous rock mass identification and extraction rules taking into account multi-dimensional features according to claim 1 is characterized in that: In S301, the slope calculation formula is as follows: ； in, Indicates the slope value measured in degrees. , Represent the increments in the horizontal and vertical directions respectively, Yes The value obtained by truncating the calculation result; In S302, the inclination angle calculation formula is as follows: ； In the formula, is the projection length of the external normal vector of the structural surface on the axis, ; In S303, it is assumed that the point to be measured in the point cloud data is ,by is the center of the sphere and the radius is The neighborhood point set in the ball is , then the point to be tested The covariance matrix of The calculation formula is: ； ； ； in, For this neighborhood The coordinates of the points, For the The number of points in the neighborhood of a point, is the centroid of the neighborhood; In S304, the three eigenvalues ​​are , the eigenvectors corresponding to the three eigenvalues ​​are , point cloud plane characteristics The formula is: ； In S305, the total variance The formula is: ； In S306, verticality The formula is: ； S307, the judgment condition is: when calculating the local features of the point cloud, the size of the spherical neighborhood search radius r is considered. If the search radius r is small, the number of points involved in the calculation will be small. If the search radius r is large, the data calculation time will be too long.

3. A system for constructing a dangerous rock mass identification and extraction rule taking into account multi-dimensional features according to any one of claims 1 to 2, characterized in that: include Data processing and algorithm improvement module: Based on the original 3D laser point cloud data, an improved progressive encryption triangulation filtering algorithm and moving curve fitting filtering algorithm are proposed to filter the vegetation. Combined with high-resolution images, a human-computer interactive method is used to check and correct the denoised point cloud to obtain the real ground point cloud; Interpretation sign construction module taking into account the development characteristics of dangerous rock masses: Based on the development characteristics of dangerous rock masses, interpretation signs for identifying dangerous rock masses taking into account the development characteristics are established from different perspectives; The module for comparing and selecting the characteristic scales of dangerous rock mass point cloud and determining its thresholds and establishing the rules for multi-scale identification and extraction of dangerous rock mass is as follows: the optimal thresholds for the inclination, planarity, total variance and verticality characteristics of dangerous rock mass are calculated and selected from different scales, and based on this, the rules for multi-scale identification and extraction of dangerous rock mass are established.

Images