Rock mass structural surface intelligent identification method and system based on three-dimensional laser scanning

Abstract

The invention provides a rock mass structural surface intelligent identification method and system based on three-dimensional laser scanning, and relates to the technical field of geological engineering, and the method comprises the steps: collecting point cloud data of a rock mass surface, and carrying out the multi-scale filtering processing based on the density distribution of the point cloud data, and obtaining multi-scale features; inputting the point cloud data subjected to the multi-scale filtering processing and the multi-scale features of the point cloud data into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural surface; on the basis of the preliminary classification result, performing plane fitting in the point cloud data by applying an RANSAC algorithm and combining multi-scale features; for the plane obtained by fitting each time, selecting a seed point from the plurality of inner points, and gradually adding adjacent points meeting a growth criterion to the current plane from the seed point by using a region growth method in combination with the growth criterion defined based on multi-scale features, so as to perfect the boundary of the rock mass structural plane; and extracting occurrence parameters, spacing and ductility of the rock mass structural plane from the plane processed by the region growing method.

Description

Technical Field

[0001] The present invention relates to the technical field of geological engineering, and particularly to an intelligent recognition method and system for rock mass structural planes based on three-dimensional laser scanning. Background Art

[0002] In the fields of geological engineering, geotechnical engineering, and mine exploitation, the accurate recognition and analysis of rock mass structural planes are crucial. Rock mass structural planes, such as joints, faults, and fissures, are important discontinuity surfaces in rock masses, and their existence has a significant impact on the stability, mechanical properties, and permeability of rock masses. Traditional methods for recognizing rock mass structural planes mainly rely on on-site manual surveys, geological mapping, and simple measurement tools. These methods are not only time-consuming and laborious but also greatly affected by human factors, making it difficult to achieve comprehensive, accurate, and rapid recognition of rock mass structural planes.

[0003] With the rapid development of three-dimensional laser scanning technology, its application in the fields of geological engineering and geotechnical engineering is becoming increasingly widespread. Three-dimensional laser scanning technology can quickly obtain the three-dimensional coordinate information of the surface of a target object by emitting laser beams and receiving the reflected signals, generating high-precision point cloud data. These point cloud data contain rich spatial geometric information and surface texture features, providing the possibility for the automatic recognition of rock mass structural planes.

[0004] However, directly recognizing rock mass structural planes from the original point cloud data still faces many challenges. First, the point cloud data on the rock mass surface often contains a large amount of noise and redundant information, such as vegetation coverage and the roughness of the rock surface, which will interfere with the recognition of structural planes. Second, the morphological and distribution characteristics of rock mass structural planes are complex and diverse, including various forms such as planes, curved surfaces, intersections, and parallels, and the performance of structural planes at different scales in the point cloud data is also different, which increases the difficulty of recognition. In addition, the recognition of rock mass structural planes also needs to consider key information such as their attitude parameters (such as dip angle, dip direction, and strike), spacing, and ductility, which are crucial for evaluating the stability of rock masses and for engineering design.

[0005] Therefore, it is necessary to provide an intelligent recognition method and system for rock mass structural planes based on three-dimensional laser scanning to solve the above technical problems. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides an intelligent recognition method and system for rock mass structural planes based on three-dimensional laser scanning, improving the recognition accuracy and efficiency of rock mass structural planes.

[0007] The present invention provides an intelligent recognition method for rock mass structural planes based on three-dimensional laser scanning, and the method includes the following steps:

[0008] Collect the point cloud data of the rock mass surface, and perform multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features;

[0009] Input the point cloud data after multi-scale filtering processing and its multi-scale features into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural plane, where the classification model is a deep learning model trained with a point cloud data sample set containing various rock mass structural plane features;

[0010] On the basis of the preliminary classification result, apply the RANSAC algorithm and perform plane fitting in the point cloud data in combination with the multi-scale features, where each plane obtained by fitting contains multiple inliers;

[0011] For each plane obtained by fitting, select seed points among the multiple inliers, and in combination with the growth criterion defined based on the multi-scale features, use the region growing method to gradually add neighboring points that meet the growth criterion to the current plane starting from the seed points to improve the boundary of the rock mass structural plane;

[0012] Extract the attitude parameters, spacing, and ductility of the rock mass structural plane from the plane processed by the region growing method.

[0013] Preferably, the collecting the point cloud data of the rock mass surface and performing multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features includes:

[0014] Calculate the local density of each point in the point cloud data to determine the density distribution of the point cloud data;

[0015] According to the density distribution of the point cloud data, determine filtering parameters of different scales, where the filtering parameters include the filtering window size and threshold;

[0016] Apply the multi-scale Gaussian filtering method to process the point cloud data according to the determined filtering parameters to extract feature points at different scales;

[0017] Fuse the feature points at different scales to form multi-scale features.

[0018] Preferably, the inputting the point cloud data after multi-scale filtering processing and its multi-scale features into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural plane includes:

[0019] Combine the spatial position of the point cloud data and the multi-scale features to construct an enhanced feature vector;

[0020] Input the point cloud data and the enhanced feature vector into the pre-trained classification model for classification inference, output the classification label of each point, and identify the points belonging to the rock mass structural plane;

[0021] Post-process the points identified as belonging to the rock mass structural plane to form a preliminary classification result of the rock mass structural plane.

[0022] Preferably, constructing an enhanced feature vector by combining the spatial position of the point cloud data and the multi-scale features includes:

[0023] Extract the spatial position features of each point from the point cloud data, where the spatial position features are three-dimensional coordinates;

[0024] Align the spatial position features of each point and the multi-scale features;

[0025] Perform feature splicing on the aligned spatial position features of each point and the multi-scale features to form an enhanced feature vector.

[0026] Preferably, based on the preliminary classification result, apply the RANSAC algorithm and perform plane fitting in the point cloud data in combination with the multi-scale features, where each plane obtained by fitting contains multiple inliers, including:

[0027] Extract the points identified as belonging to the rock mass structural plane from the preliminary classification result to form a candidate point set, and perform feature enhancement processing in combination with the corresponding multi-scale features;

[0028] In the candidate point set, apply the RANSAC algorithm for plane fitting, where the RANSAC algorithm randomly selects a set of points from the candidate point set in an iterative manner to fit a plane and evaluates the matching degree of the plane with other points in the candidate point set;

[0029] In each iteration, combine the multi-scale features to evaluate the quality of the plane obtained by fitting;

[0030] Set an inlier threshold and obtain the inliers belonging to the plane obtained by fitting, where the inliers refer to the points whose distance from the plane obtained by fitting is less than the inlier threshold;

[0031] After each iteration, evaluate the quality of the plane obtained by fitting in the current iteration according to the number and quality of the inliers, and select the optimal plane obtained by fitting as the result of the current iteration.

[0032] Preferably, for each plane obtained by fitting, select seed points among the multiple inliers, and combine the growth criterion defined based on the multi-scale features, and use the region growing method to gradually add neighboring points that meet the growth criterion to the current plane starting from the seed points to improve the boundary of the rock mass structural plane, including:

[0033] Select one or more points from the multiple inliers included in each plane obtained by fitting as seed points;

[0034] Based on multi-scale features, a growth criterion is defined to determine whether neighboring points should be added to the current plane, where the growth criterion includes the distance between the neighboring points and the seed point, the consistency of the normal direction, and the similarity of the multi-scale features;

[0035] Based on the selected seed points, the region growing method is used to gradually add neighboring points that meet the growth criterion to the current plane, and during the process of adding each point, the matching degree between the newly added point and the current plane is re-evaluated;

[0036] Through multiple iterations and growths until no more neighboring points that meet the growth criterion can be found.

[0037] Preferably, extracting the attitude parameters, spacing, and ductility of the rock mass structural plane from the plane processed by the region growing method includes:

[0038] Smoothing the plane processed by the region growing method;

[0039] Using the smoothed plane to calculate the attitude parameters of the rock mass structural plane, where the attitude parameters include dip angle, dip direction, and strike;

[0040] Between the confirmed rock mass structural planes, measure the spacing between adjacent structural planes, where the spacing includes vertical spacing and horizontal spacing;

[0041] Obtain the ductility of the rock mass structural plane by calculating the length, width of the rock mass structural plane, and the connection parameters with other structural planes.

[0042] The present invention also provides an intelligent recognition system for rock mass structural planes based on 3D laser scanning, which is used to execute the intelligent recognition method for rock mass structural planes based on 3D laser scanning. The system includes:

[0043] A point cloud data multi-scale filtering module, which is used to collect the point cloud data of the rock mass surface and perform multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features;

[0044] A preliminary classification module for rock mass structural planes, which is used to input the point cloud data processed by multi-scale filtering and its multi-scale features into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural plane, where the classification model is a deep learning model trained by a point cloud data sample set containing various rock mass structural plane features;

[0045] A RANSAC plane fitting module, which is used to apply the RANSAC algorithm on the basis of the preliminary classification result and perform plane fitting in the point cloud data in combination with the multi-scale features, where each plane obtained by fitting contains multiple inliers;

[0046] The regional growth method boundary improvement module is used to select seed points from the multiple inliers for each plane obtained by fitting, and in combination with the growth criterion defined based on multi-scale features, use the regional growth method to gradually add adjacent points that meet the growth criterion to the current plane starting from the seed points to improve the boundary of the rock mass structural plane;

[0047] The structural plane parameter extraction module is used to extract the attitude parameters, spacing, and ductility of the rock mass structural plane from the plane processed by the regional growth method.

[0048] Compared with the related technologies, the intelligent recognition method and system of rock mass structural plane based on three-dimensional laser scanning provided by the present invention have the following beneficial effects:

[0049] The present invention collects the point cloud data on the surface of the rock mass and performs multi-scale filtering processing to extract the feature information at different scales. Subsequently, a pre-trained deep learning model is used to preliminarily classify the point cloud data, and in combination with the RANSAC algorithm and the regional growth method, the recognition result is further refined, and finally the key parameters of the rock mass structural plane are extracted. This method can not only effectively remove noise and redundant information, improve the accuracy and efficiency of recognition, but also comprehensively and accurately obtain the morphology and distribution characteristics of the rock mass structural plane, providing strong technical support for the fields of geological engineering and geotechnical engineering. Brief Description of the Drawings

[0050] Figure 1 It is a flowchart of the intelligent recognition method of rock mass structural plane based on three-dimensional laser scanning provided by the present invention;

[0051] Figure 2 It is a module structure diagram of the intelligent recognition system of rock mass structural plane based on three-dimensional laser scanning provided by the present invention. Detailed Embodiments

[0052] The present invention will be further described in detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. In addition, it should be noted that for the convenience of description, only parts related to the present invention are shown in the drawings, not all structures. In addition, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0053] It should also be noted that, for ease of description, only the parts related to the present invention rather than all the content are shown in the drawings. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. When the operations are completed, the process can be terminated, but there can also be additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0054] Embodiment 1

[0055] The present invention provides an intelligent recognition method for rock mass discontinuity surfaces based on three-dimensional laser scanning. Referring to Figure 1 as shown, referring to Figure 1 as shown, the method includes the following steps:

[0056] S1: Collect the point cloud data of the rock mass surface, and perform multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features.

[0057] Specifically, step S1 includes the following steps:

[0058] S11: Calculate the local density of each point in the point cloud data to determine the density distribution of the point cloud data.

[0059] In three-dimensional laser scanning technology, the point cloud data of the rock mass surface obtained usually has an uneven density distribution, which may be caused by scanning angles, distances, or object surface characteristics. To effectively identify and analyze rock mass discontinuity surfaces, it is necessary to first understand the density characteristics of the point cloud data. For this purpose, in step S11, the local density around each point is calculated.

[0060] This process can be achieved by counting the number of points in the neighborhood of each point. The neighborhood size can be defined as a spherical or cubic window, and its radius or side length is preset according to the resolution of the point cloud data and the expected scale of the discontinuity surface. By this method, a density distribution map of the entire point cloud dataset can be obtained, which can show which areas are dense and which areas are sparse in the point cloud, and this is crucial for subsequent filtering processing.

[0061] S12: Determine the filtering parameters of different scales according to the density distribution of the point cloud data, where the filtering parameters include the filtering window size and threshold.

[0062] In this embodiment, based on the density distribution information obtained in step S11, the filtering parameters can be set specifically to adapt to the data characteristics of different density regions. For high-density regions, a smaller filtering window may be sufficient because the details in these regions are more abundant; while for low-density regions, a larger filtering window is required to ensure that enough points participate in the filtering, thereby reducing the influence of noise.

[0063] In addition, a threshold is also set to distinguish which points should be retained and which points are considered noise and removed. The selection of this threshold depends on the understanding of the characteristics of the rock mass structural plane and experimental verification. The reasonable selection of filtering parameters helps to improve the accuracy and efficiency of feature extraction.

[0064] S13: Apply the multi-scale Gaussian filtering method to process the point cloud data according to the determined filtering parameters to extract feature points at different scales.

[0065] In this embodiment, once the filtering parameters are determined, the multi-scale Gaussian filtering method can be used to smooth the point cloud data and highlight the feature points at different scales. Gaussian filtering is a common noise reduction method that achieves a smoothing effect by replacing the position of the original point with the weighted average of its neighboring points.

[0066] In this process, different scales (i.e., Gaussian functions with different widths) are applied to the point cloud to capture various structural plane features from small to large. After filtering, the features originally masked by noise become more obvious, while the unnecessary details are weakened or eliminated. In this way, the geometric feature points related to the rock mass structural plane can be identified more accurately.

[0067] S14: Fuse the feature points at different scales to form multi-scale features.

[0068] In this embodiment, the last step is to establish a connection between the feature points extracted at different scales to form a multi-scale feature representation. This is because the rock mass structural plane exhibits different features at different scales. For example, cracks may have both macroscopic trends and microscopic textures. Therefore, simply selecting feature points at a single scale is not sufficient to comprehensively describe the structural plane. By fusing information from multiple scales, a more complete and robust feature representation can be constructed, which is very crucial for the subsequent classification and fitting steps. The fusion method can be to directly splice the feature vectors at each scale, or to automatically find the best combination method through a learning algorithm. The finally formed multi-scale features not only contain spatial position information but also shape and texture information at different scales, greatly improving the ability of intelligent recognition of the rock mass structural plane.

[0069] S2: Input the point cloud data and its multi-scale features that have undergone multi-scale filtering into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural plane, where the classification model is a deep learning model trained with a point cloud data sample set containing various rock mass structural plane features.

[0070] Specifically, step S2 includes the following steps:

[0071] S21: Combine the spatial position of the point cloud data and the multi-scale features to construct an enhanced feature vector.

[0072] In this embodiment, in order to enable the classification model to fully utilize the information in the point cloud data, in step S21, an enhanced feature vector is constructed. First, extract the spatial position features of each point from the point cloud data, that is, the three-dimensional coordinates (x, y, z), which provide the exact position of the point in space. Then, perform alignment processing on the spatial position features of each point and the multi-scale features obtained through step S1. The alignment process ensures the correct association between the spatial position and the multi-scale features, that is, the position information of each point is matched with its corresponding multi-scale features. Finally, perform feature splicing on the aligned spatial position features and multi-scale features to form an enhanced feature vector. This vector contains the spatial position of the point and its features such as shape and texture at different scales, thus providing rich information for subsequent classification tasks.

[0073] More specifically, step S21 includes the following steps:

[0074] a. Extract the spatial position features of each point from the point cloud data, where the spatial position features are three-dimensional coordinates.

[0075] In this embodiment, extract the spatial position features of each point from the point cloud data, and these features are represented as three-dimensional coordinates. The spatial position features are crucial for understanding the point cloud data because they directly determine the exact position of the point in three-dimensional space. The three-dimensional coordinates are the most basic components of the point cloud data and are also the basis for other advanced analyses. In this step, traverse the entire point cloud data set and record the three-dimensional coordinates for each point. The technical effect of this step is to provide accurate position information for subsequent processing, enabling us to further analyze and process the points based on geometric relationships.

[0076] b. Perform alignment processing on the spatial position features of each point and the multi-scale features.

[0077] In this embodiment, the spatial position features of each point and the multi-scale features obtained through step S1 are aligned. The purpose of the alignment process is to ensure the correct association between the spatial position and the multi-scale features. Since the point cloud data may be irregularly distributed and the multi-scale features are calculated independently at different scales, a process is required to ensure that the multi-scale features of each point correspond to its spatial position features. Algorithms such as nearest neighbor search can be used to find the multi-scale features corresponding to the closest spatial position features. Through the alignment process, the accuracy of subsequent stitching operations can be ensured, thus avoiding classification errors caused by feature mismatches.

[0078] c. Feature stitching is performed on the spatial position features and the multi-scale features of each aligned point to form an enhanced feature vector.

[0079] In this embodiment, the spatial position features and multi-scale features of each aligned point are stitched to form an enhanced feature vector. Feature stitching refers to combining two or more different types of features into a new feature representation. Here, the three-dimensional coordinates (spatial position features) of each point are merged with the shape and texture features (multi-scale features) of the point at different scales to create a more powerful feature vector. This enhanced feature vector not only contains the original geometric information but also incorporates the structured information obtained through filtering, greatly enhancing the expressive power of the features. Such an enhanced feature vector can provide more context information to the deep learning classification model, helping to improve the model's understanding and recognition accuracy of complex rock mass structural plane features.

[0080] S22: The point cloud data and the enhanced feature vector are input into a pre-trained classification model for classification inference, and the classification label of each point is output, and the points belonging to the rock mass structural plane are identified.

[0081] In this embodiment, the point cloud data and the enhanced feature vector are input into a pre-trained deep learning classification model. This classification model is trained on a large number of point cloud data sample sets containing various rock mass structural plane features, so it has the ability to identify different types of rock mass structural planes. In this process, the classification model will perform complex calculations and inferences based on the input point cloud data and enhanced feature vector, and output the classification label of each point. The classification label is used to identify which points belong to the rock mass structural plane and which do not. The result of this step is a preliminary classification result, in which each point is marked as possibly belonging or not belonging to a part of the rock mass structural plane.

[0082] S23: Post-processing is performed on the identified points belonging to the rock mass structural plane to form a preliminary classification result of the rock mass structural plane.

[0083] In this embodiment, post - processing is performed on the points identified as belonging to the rock mass structural plane to form a more accurate and coherent preliminary classification result of the rock mass structural plane. This generally involves operations such as removing isolated points, filling small holes, and connecting disconnected structural plane segments. The purpose is to improve the quality of the classification result, make the boundary of the structural plane clearer and smoother, and reduce the possibility of misclassification.

[0084] In addition, morphological operations such as dilation and erosion can also be applied to optimize the shape of the structural plane. Through these post - processing operations, a geometric shape closer to the actual rock mass structural plane can be obtained, providing a reliable basis for subsequent plane fitting and other analyses.

[0085] S3: Based on the preliminary classification result, apply the RANSAC algorithm and combine the multi - scale features to perform plane fitting in the point cloud data, where each plane obtained by fitting contains multiple inliers.

[0086] Specifically, step S3 includes the following steps:

[0087] S31: Extract the points identified as belonging to the rock mass structural plane from the preliminary classification result to form a candidate point set, and perform feature enhancement processing in combination with the corresponding multi - scale features.

[0088] In this embodiment, the preliminary classification result has roughly classified the point cloud data through a deep learning model, but these classifications may contain noisy or misclassified points. To improve the accuracy of subsequent plane fitting, a cleaner point set is needed as a basis.

[0089] Therefore, the task of this step is to screen out the points most likely to belong to the rock mass structural plane from the preliminary classification result. At the same time, use the multi - scale features obtained in step S1 to perform feature enhancement processing on these points to ensure that each point not only contains its original spatial position information but also integrates shape and texture features at different scales. The technical effect of this step is to provide a high - quality candidate point set, providing a reliable data basis for the subsequent plane fitting.

[0090] S32: In the candidate point set, apply the RANSAC algorithm for plane fitting, where the RANSAC algorithm randomly selects a set of points from the candidate point set in an iterative manner to fit a plane and evaluates the matching degree of this plane with other points in the candidate point set.

[0091] In this embodiment, RANSAC is an iterative algorithm used to estimate the parameters of a mathematical model, especially when the data contains a large number of outliers. In this step, the RANSAC algorithm randomly selects a set of points from the candidate point set to attempt to fit a plane, and then evaluates the matching degree of this plane with other points in the candidate point set. This process is iterative, and a new plane hypothesis is generated in each iteration, and the fitness of this plane to all points is calculated.

[0092] Specifically, RANSAC will select a certain number of points (such as three non - collinear points) to define a plane, and then calculate the distances from the remaining points to this plane to determine which points can be regarded as "inliers", that is, points that conform to the plane hypothesis. Through multiple iterations, RANSAC can find the optimal plane model, even if there are a large number of noises or outliers in the data. The technical effect of this step is to find the plane model that best represents the rock mass structural plane through iterative optimization, improving the accuracy and robustness of the fitting result.

[0093] S33: In each iteration, combine multi - scale features to evaluate the quality of the fitted plane.

[0094] In this embodiment, in each iteration, combine multi - scale features to evaluate the quality of the fitted plane. Since the rock mass structural plane may exhibit different characteristics at different scales, when evaluating the plane quality, in addition to considering the distance from points to the plane, it is also necessary to combine multi - scale features to judge the rationality of the plane.

[0095] For example, it can be checked whether the points on the fitted plane have similar multi - scale features, or whether the direction of the plane is consistent with the normal direction of the surrounding points. This comprehensive evaluation method can more accurately identify the plane that conforms to the characteristics of the rock mass structural plane, avoiding misjudgment caused by single geometric features.

[0096] S34: Set an inlier threshold and obtain the inliers that belong to the fitted plane, where the inliers refer to the points whose distance from the fitted plane is less than the inlier threshold.

[0097] In this embodiment, set an inlier threshold and obtain the inliers that belong to the fitted plane. Inliers refer to those points whose distance from the fitted plane is less than the set threshold. The selection of this threshold depends on the understanding of the characteristics of the rock mass structural plane and experimental verification, and usually needs to be adjusted according to specific project requirements.

[0098] By setting a reasonable inlier threshold, points that do not conform to the plane hypothesis can be effectively filtered out, thereby improving the purity of the fitting result.

[0099] S35: After each iteration, evaluate the quality of the currently fitted plane based on the quantity and quality of the inliers, and select the optimal fitted plane as the result of the current iteration.

[0100] In this embodiment, the evaluation criteria include the quantity of inliers, the uniformity of inlier distribution, the smoothness of the plane, etc. By comparing the fitting results in different iterations, select the plane that best meets the characteristics of the rock mass structural plane as the final fitting result. The technical effect of this step is to ensure that the final plane fitting result is not only geometrically reasonable but also closest to the actual rock mass structural plane in terms of feature expression, providing a reliable basis for subsequent analysis.

[0101] S4: For each plane obtained by fitting, select seed points from the multiple inliers, and in combination with the growth criterion defined based on multi-scale features, use the region growing method to gradually add neighboring points that meet the growth criterion to the current plane starting from the seed points to improve the boundary of the rock mass structural plane.

[0102] Specifically, step S4 includes the following steps:

[0103] S41: Select one or more points from the multiple inliers included in each plane obtained by fitting as seed points.

[0104] In this embodiment, the selection of seed points is crucial for the subsequent region growing process because they are the starting points of growth. Ideal seed points should be located on the rock mass structural plane and have a high confidence level, that is, the probability that these points are correctly classified as the structural plane is relatively high.

[0105] To this end, those points that are inliers in the RANSAC algorithm and have the highest degree of matching with the fitted plane can be selected as seed points. In addition, multiple seed points distributed at different positions can also be considered to ensure that the growth process can cover the entire structural plane.

[0106] S42: Based on multi-scale features, define a growth criterion for determining whether neighboring points should be added to the current plane, where the growth criterion includes the distance between neighboring points and seed points, the consistency of the normal direction, and the similarity of multi-scale features.

[0107] In this embodiment, a growth criterion for determining whether neighboring points should be added to the current plane is defined based on multi-scale features. The growth criterion is the core of the region growing method, which determines which neighboring points can be regarded as part of the structural plane and added to the current plane. The growth criterion generally includes the following aspects:

[0108] The distance between neighboring points and seed points: The closer the distance between points, the more likely they belong to the same structural plane. Therefore, a maximum distance threshold can be set, and points exceeding this threshold will not be considered.

[0109] Consistency of normal direction: Rock mass structural planes usually have relatively consistent normal directions. Therefore, if the normal direction of an adjacent point is similar to that of the seed point or the current plane, it is more likely that the point belongs to the same structural plane. The consistency can be evaluated by calculating the angle between the two normals, and an angle threshold can be set.

[0110] Similarity of multi-scale features: Since rock mass structural planes exhibit different features at different scales, the similarity of multi-scale features is also an important criterion for judgment. For example, if the shape and texture features of an adjacent point are similar to those of the seed point at multiple scales, it is more likely that the point belongs to the same structural plane. This can be evaluated by comparing the distance or similarity between multi-scale feature vectors.

[0111] By comprehensively considering the above three aspects of factors, a set of strict growth criteria can be defined to ensure that only the points that truly belong to the structural plane will be added to the current plane.

[0112] S43: Based on the selected seed point, use the region growing method to gradually add adjacent points that meet the growth criteria to the current plane, and during the process of adding each point, re-evaluate the matching degree between the newly added point and the current plane.

[0113] In this embodiment, based on the selected seed point, search for its adjacent points layer by layer, and judge whether these adjacent points should be added to the current plane according to the predefined growth criteria. After each new point is added, recalculate the parameters of the current plane (such as normal direction, center point, etc.), and evaluate the matching degree between the newly added point and the updated plane. If an adjacent point meets the growth criteria and has a high matching degree with the current plane, it is added to the current plane; otherwise, the point is not added. This process is iterative until no more adjacent points meet the growth criteria. By gradually expanding, the boundary of the rock mass structural plane is gradually improved while maintaining the consistency of the geometric characteristics and multi-scale features of the structural plane.

[0114] S44: Through multiple iterations and growth until no more adjacent points that meet the growth criteria can be found.

[0115] In this embodiment, through multiple iterations and growth until no more neighboring points that meet the growth criteria can be found. During this process, as more and more points are added to the current plane, the boundary of the structural plane gradually becomes complete and clear. To ensure that the termination conditions of the growth process are reasonable, some additional termination conditions are set, including but not limited to the maximum number of iterations or the maximum growth range. When these conditions are reached, even if there are still points that meet the growth criteria, the growth process can be stopped. This ensures that the growth process neither expands excessively nor terminates prematurely, thus obtaining a complete and accurate boundary of the rock mass structural plane.

[0116] S5: Extract the attitude parameters, spacing, and ductility of the rock mass structural plane from the plane processed by the region growing method.

[0117] Specifically, step S5 includes the following steps:

[0118] S51: Smooth the plane processed by the region growing method.

[0119] In this embodiment, since the region growing method may introduce some small geometric irregularities or noises, especially in the boundary region, before calculating the attitude parameters, it is necessary to smooth these planes to improve the data quality. The smoothing process can reduce the tiny fluctuations on the surface, making the subsequent parameter calculation more accurate and stable. The smoothing methods include but are not limited to moving average filtering or bilateral filtering.

[0120] S52: Use the smoothed plane to calculate the attitude parameters of the rock mass structural plane, where the attitude parameters include dip angle, dip direction, and strike.

[0121] In this embodiment, the attitude parameters are important indicators to describe the spatial attitude of the rock mass structural plane and are crucial for understanding the mechanical properties and stability of the rock mass. The specific calculation process is as follows:

[0122] Dip angle: It refers to the angle between the structural plane and the horizontal plane. It can be determined by calculating the angle between the normal vector of each point and the vertical direction. For a given plane, its dip angle is a constant.

[0123] Dip direction: It refers to the direction in which the structural plane slopes downward, usually expressed in angles, ranging from 0° to 360°. It can be determined by calculating the projection direction of the normal vector of the structural plane on the horizontal plane.

[0124] Strike: It refers to the projection direction of the intersection line of the structural plane on the horizontal plane, usually also expressed in angles. It can be determined by calculating the projection direction of the perpendicular line of the normal vector of the structural plane on the horizontal plane.

[0125] S53: Measure the spacing between adjacent rock mass structural planes among the identified rock mass structural planes, where the spacing includes vertical spacing and horizontal spacing.

[0126] In this embodiment, it is necessary to measure the spacing between adjacent structural planes, including vertical spacing and horizontal spacing. The vertical spacing refers to the shortest distance between two parallel structural planes, usually measured along the direction perpendicular to the structural plane; the horizontal spacing refers to the shortest distance between two structural planes in the horizontal direction.

[0127] To accurately measure these spacings, algorithms including but not limited to the nearest neighbor search algorithm can be used. The measurement results can help understand the distribution density and arrangement pattern of the structural planes, which are very useful for evaluating the overall stability of the rock mass and potential sliding risks.

[0128] S54: Obtain the ductility of the rock mass structural planes by calculating the length, width of the rock mass structural planes and the connection parameters with other structural planes.

[0129] In this embodiment, the length and width of the structural plane can be directly measured from its boundary, while the connection parameters reflect the mutual relationship between the structural plane and other structural planes. For example, if two structural planes intersect, the length and angle of their intersection line are also important connection parameters. Through these calculations, the morphological characteristics of the structural plane and its position and role in the network can be comprehensively understood.

[0130] Embodiment 2

[0131] The present invention also provides an intelligent identification system for rock mass structural planes based on three-dimensional laser scanning, which is used to execute the intelligent identification method for rock mass structural planes based on three-dimensional laser scanning. As shown in Figure 2 The system includes:

[0132] A point cloud data multi-scale filtering module 100, which is used to collect the point cloud data on the surface of the rock mass and perform multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features.

[0133] A preliminary classification module 200 for rock mass structural planes, which is used to input the point cloud data and its multi-scale features after multi-scale filtering processing into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural planes, where the classification model is a deep learning model trained with a point cloud data sample set containing various rock mass structural plane features.

[0134] A RANSAC plane fitting module 300, which is used to apply the RANSAC algorithm based on the preliminary classification result and perform plane fitting in the point cloud data in combination with the multi-scale features, where each plane obtained by fitting contains multiple inliers.

[0135] The region growing method boundary refinement module 400 is used to select seed points from the multiple inliers for each fitted plane, and in combination with the growth criterion defined based on multi-scale features, use the region growing method to gradually add neighboring points that meet the growth criterion to the current plane starting from the seed points, so as to refine the boundary of the rock mass structural plane.

[0136] The structural plane parameter extraction module 500 is used to extract the attitude parameters, spacing, and ductility of the rock mass structural plane from the plane processed by the region growing method.

[0137] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each process and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in one Figure 1 process or multiple processes and / or blocks Figure 1 block or multiple blocks.

[0138] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable storage medium, and the storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0139] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, commodity or device comprising the element.

Claims

1. An intelligent identification method of rock mass structural surface based on three-dimensional laser scanning, characterized in that: The method comprises the following steps: Collecting point cloud data of the rock mass surface, and performing multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features; Inputting the point cloud data processed by multi-scale filtering and its multi-scale features into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural surface, wherein the classification model is a deep learning model trained with a point cloud data sample set containing multiple rock mass structural surface features; Based on the preliminary classification results, applying the RANSAC algorithm and combining the multi-scale features to perform plane fitting in the point cloud data, wherein each fitted plane contains a plurality of interior points; For each plane obtained by fitting, a seed point is selected from the multiple internal points, and a growth criterion defined based on multi-scale features is combined, and a regional growing method is used to gradually add neighboring points that meet the growth criterion to the current plane starting from the seed point to improve the boundary of the rock mass structural surface; The occurrence parameters, spacing and ductility of the rock mass structural planes are extracted from the plane processed by the region growing method.

2. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 1 is characterized in that: The point cloud data of the rock mass surface is collected, and multi-scale filtering is performed based on the density distribution of the point cloud data to obtain multi-scale features, including: Calculating the local density of each point in the point cloud data to determine the density distribution of the point cloud data; Determining filtering parameters of different scales according to the density distribution of the point cloud data, wherein the filtering parameters include a filtering window size and a threshold; Applying a multi-scale Gaussian filtering method to process the point cloud data according to the determined filtering parameters to extract feature points at different scales; The feature points at different scales are fused to form multi-scale features.

3. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 2 is characterized in that: The point cloud data processed by multi-scale filtering and its multi-scale features are input into the pre-trained classification model to obtain the preliminary classification results of the rock mass structural surface, including: Combining the spatial position of the point cloud data and the multi-scale features, constructing an enhanced feature vector; Inputting the point cloud data and the enhanced feature vector into a pre-trained classification model, performing classification reasoning, outputting a classification label for each point, and identifying points belonging to a rock mass structural surface; The marked points belonging to the rock mass structural surface are post-processed to form the preliminary classification results of the rock mass structural surface.

4. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 3 is characterized in that: The step of combining the spatial position of the point cloud data and the multi-scale features to construct an enhanced feature vector comprises: Extracting a spatial position feature of each point from the point cloud data, wherein the spatial position feature is a three-dimensional coordinate; Aligning the spatial position feature of each point with the multi-scale feature; The spatial position feature of each aligned point and the multi-scale feature are concatenated to form an enhanced feature vector.

5. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 4 is characterized in that: Based on the preliminary classification results, the RANSAC algorithm is applied, and the multi-scale features are combined to perform plane fitting on the point cloud data, wherein each fitted plane contains multiple internal points, including: Extracting points identified as belonging to rock mass structural surfaces from the preliminary classification results to form a candidate point set, and performing feature enhancement processing in combination with the corresponding multi-scale features; In the candidate point set, a RANSAC algorithm is applied to fit a plane, wherein the RANSAC algorithm randomly selects a set of points from the candidate point set in an iterative manner to fit a plane, and evaluates the matching degree of the plane with other points in the candidate point set; In each iteration, multi-scale features are combined to evaluate the quality of the fitted plane; Setting an inlier threshold and obtaining inliers in the fitted plane, wherein the inliers refer to points whose distance from the fitted plane is less than the inlier threshold; After each iteration, the quality of the currently fitted plane is evaluated according to the number and quality of the interior points, and the best fitted plane is selected as the result of the current iteration.

6. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 5, characterized in that: For each plane obtained by fitting, a seed point is selected from the multiple interior points, and a growth criterion defined based on multi-scale features is combined, and a regional growing method is used to gradually add neighboring points that meet the growth criterion to the current plane from the seed point to improve the boundary of the rock mass structural surface, including: Select one or more points as seed points from the multiple internal points contained in the plane obtained by each fitting; Based on the multi-scale features, a growth criterion for determining whether a neighboring point should be added to the current plane is defined, wherein the growth criterion includes a distance between the neighboring point and the seed point, consistency of the normal direction, and similarity of the multi-scale features; Based on the selected seed point, the region growing method is used to gradually add neighboring points that meet the growth criteria to the current plane, and in the process of adding points each time, the matching degree between the newly added points and the current plane is re-evaluated; Through multiple iterations and growth, until no neighboring points that meet the growth criteria can be found.

7. The method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning according to claim 6, characterized in that: The method of extracting the occurrence parameters, spacing and ductility of the rock mass structural plane from the plane processed by the region growing method includes: Smoothing the plane processed by the region growing method; Utilizing the smoothed plane, calculating the occurrence parameters of the rock mass structural surface, wherein the occurrence parameters include dip, inclination and strike; Between the confirmed rock mass structural planes, measuring the distance between adjacent structural planes, wherein the distance includes vertical distance and horizontal distance; The ductility of the rock mass structural surface is obtained by calculating the length, width and connection parameters of the rock mass structural surface with other structural surfaces.

8. A system for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning, used to execute the method for intelligently identifying rock mass structural surfaces based on three-dimensional laser scanning as claimed in any one of claims 1 to 7, characterized in that: The system comprises: A point cloud data multi-scale filtering module is used to collect point cloud data on the surface of the rock mass, and perform multi-scale filtering processing based on the density distribution of the point cloud data to obtain multi-scale features; A rock mass structural surface preliminary classification module is used to input the point cloud data processed by multi-scale filtering and its multi-scale features into a pre-trained classification model to obtain a preliminary classification result of the rock mass structural surface, wherein the classification model is a deep learning model trained with a point cloud data sample set containing multiple rock mass structural surface features; A RANSAC plane fitting module, for applying the RANSAC algorithm based on the preliminary classification result and performing plane fitting in the point cloud data in combination with the multi-scale features, wherein each fitted plane contains a plurality of interior points; A regional growing method boundary improvement module is used to select a seed point from the multiple internal points for each plane obtained by fitting, and use the regional growing method to gradually add neighboring points that meet the growth criteria to the current plane starting from the seed point in combination with the growth criteria defined based on multi-scale features, so as to improve the boundary of the rock mass structural surface; The structural surface parameter extraction module is used to extract the occurrence parameters, spacing and ductility of the rock mass structural surface from the plane processed by the regional growing method.

Images