Mine modeling method and system based on point cloud data, terminal and medium

Abstract

The invention relates to the field of digital modeling, and particularly discloses a mine modeling method and system based on point cloud data, a terminal and a medium, and the method comprises the steps: obtaining original point cloud data of a target mine area; preprocessing the original point cloud data, including voxel filtering, to obtain target point cloud data; calculating the curvature and the normal vector of each point in the target point cloud data; inputting the coordinate, the curvature and the normal vector of each point as input values into a pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories; constructing an energy function of a graph cut algorithm based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data, and obtaining the final geological category to which each point belongs through the graph cut algorithm; and constructing a digital model of the target mine area based on the geological category to which each point belongs. According to the method, geological category prediction is carried out based on the multi-scale features of the point cloud, optimization is carried out in combination with the graph cut algorithm, and the classification efficiency and robustness are improved.

Description

Technical Field

[0001] The present invention relates to the field of digital modeling, and in particular to a mine modeling method, system, terminal and medium based on point cloud data. Background Art

[0002] In the field of mine geological modeling, traditional mine modeling methods primarily rely on manual geological surveys, borehole data interpolation, or total station measurements. These methods suffer from low efficiency, high cost, and difficulty characterizing complex geological structures (such as faults and fracture zones). In recent years, 3D laser scanning and photogrammetry technologies have enabled the rapid acquisition of mine point cloud data. Point cloud data can densely and comprehensively capture spatial information about a mining area, providing a rich data foundation for mine modeling.

[0003] However, raw point cloud data contains a significant amount of noise and redundant information, and using it directly for modeling can severely impact accuracy and reliability. Furthermore, methods often rely solely on point cloud coordinate information, ignoring geometric features like curvature and normal vectors. This results in low accuracy in identifying key geological features like veins and faults. Furthermore, existing deep learning models (such as PointNet) directly output hard classification labels, lacking the ability to coordinate local geometric details with global semantic associations. This makes it difficult to handle point cloud sparsity and noise interference, resulting in poor classification robustness. Summary of the Invention

[0004] To solve the above problems, the present invention provides a mine modeling method, system, terminal and medium based on point cloud data. Voxel filtering is first performed to reduce data redundancy, and then geological category prediction is performed based on the multi-scale characteristics of the point cloud. The system is optimized in combination with the graph cut algorithm to achieve improved classification efficiency and robustness.

[0005] In a first aspect, the technical solution of the present invention provides a mine modeling method based on point cloud data, comprising the following steps: Obtain original point cloud data of the target mining area; Preprocess the original point cloud data, including voxel filtering, to obtain the target point cloud data; Calculate the curvature and normal vector of each point in the target point cloud data; The coordinates, curvature, and normal vector of each point are input into a pre-trained geological classification model to obtain the probability that each point belongs to various geological categories; Based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, the energy function of the graph cut algorithm is constructed, and the final geological category of each point is obtained through the graph cut algorithm; A digital model of the target mining area is constructed based on the geological category to which each point belongs.

[0006] In an optional embodiment, an energy function of a graph cut algorithm is constructed based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data. The final geological category to which each point belongs is obtained by the graph cut algorithm, specifically including: Based on the predicted geological category of each point, the data item of the energy function of the graph cut algorithm is constructed, and the geological category corresponding to the maximum probability of each point belonging to various geological categories output by the geological category classification model is used as the initial predicted geological category; The smoothing term of the energy function of the graph cut algorithm is constructed based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data; The predicted geological category of each point is adjusted, and the final geological category of each point is obtained with the goal of minimizing the energy function.

[0007] In an optional embodiment, the point cloud features of the target point cloud data include the spatial distance between two points and the normal vector of the point.

[0008] In an optional embodiment, the expression of the energy function is:

[0009] in, ; ; Where, For smooth number, is a data item, is the smoothing term, is the total number of points in the target point cloud, for point Belongs to the geological category to which the prediction belongs The probability of is the adjacency relationship between points, for A collection of geological categories, and Points and point Belongs to the geological category The probability of for point and point The Euclidean distance between and Points and point The normal vector of is the weight coefficient, is the scale parameter.

[0010] In an optional embodiment, the geological category classification model is constructed by a PointNet++ network.

[0011] In an optional embodiment, a digital model of the target mining area is constructed based on the geological category of each point, specifically including: Divide the points belonging to the same geological category into one region to obtain multiple segmented regions; A Poisson surface reconstruction algorithm is used to generate a smooth surface for each segmented area, which is the local model. The ICP algorithm is used to merge the local models into an overall model in a unified coordinate system; The moving least squares interpolation method is used to fill the data missing areas of the overall model to obtain a digital model of the target mining area.

[0012] In an optional embodiment, preprocessing the original point cloud data further includes statistical filtering, removing vegetation and interference points through plane fitting classification and elevation threshold method, and performing ICP algorithm registration using control points.

[0013] In a second aspect, the technical solution of the present invention provides a mine modeling system based on point cloud data, comprising: The original point cloud acquisition module is used to obtain the original point cloud data of the target mining area; Point cloud preprocessing module, used to preprocess the original point cloud data, including voxel filtering, to obtain target point cloud data; Point cloud feature calculation module, used to calculate the curvature and normal vector of each point in the target point cloud data; The geological category prediction module is used to input the coordinates, curvature, and normal vector of each point into the pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories; The geological category determination module is used to construct the energy function of the graph cut algorithm based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, and finally obtain the geological category to which each point belongs through the graph cut algorithm; The digital model building module is used to build a digital model of the target mining area based on the geological category to which each point belongs.

[0014] In a third aspect, the technical solution of the present invention provides a terminal, including: A memory for storing a mine modeling program based on point cloud data; A processor is configured to implement the steps of any of the above-mentioned methods for mine modeling based on point cloud data when executing the mine modeling program based on point cloud data.

[0015] In a fourth aspect, the technical solution of the present invention provides a computer-readable storage medium, on which a mine modeling program based on point cloud data is stored. When the mine modeling program based on point cloud data is executed by a processor, the steps of the mine modeling method based on point cloud data as described in any one of the above items are implemented.

[0016] It can be seen from the above technical solution that the present application has the following advantages: after obtaining the original point cloud data, preprocessing is performed, voxel filtering is performed during the preprocessing process, and then the curvature and normal vector of the point are calculated. The coordinates, curvature, and normal vector of each point are used as input to process the pre-trained geological category classification model to output the probability that each point belongs to various geological categories, and then the energy function of the graph cut algorithm is constructed based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data. The final geological category of each point is obtained through the graph cut algorithm, and finally a digital model of the target mining area is constructed based on the geological category to which each point belongs. The present invention uses voxel filtering to reduce the amount of redundant data in point clouds, retains key geometric structures, improves computational efficiency, combines curvature and normal vectors to construct multi-dimensional feature descriptions, and enhances local geometric representation capabilities (such as accurately identifying fault edges in high curvature areas). The end-to-end classification framework based on deep learning models (such as PointNet++) integrates local details (normal vector changes) and global context (vein direction), improves the classification accuracy of complex geological categories such as veins and fracture zones, and outputs point-by-point probability distributions rather than hard labels, which can provide flexible constraints for post-optimization. Graph cut algorithms are then introduced to jointly optimize classification probability and spatial smoothness through energy functions, correct isolated misclassified points (such as integrating discrete "vein" points into continuous ore bodies), ensure the natural transition of geological boundaries, effectively solve the sparsity of point clouds and noise interference, and improve classification robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 A schematic flow chart of a mine modeling method based on point cloud data provided by an embodiment of the present invention.

[0019] Figure 2 is the original point cloud and section diagram, where Figure 2 (a) is a schematic diagram of the original point cloud. Figure 2 (b) is a schematic diagram of the original point cloud section.

[0020] Figure 3 Schematic diagram of DEM generated based on the original point cloud.

[0021] Figure 4 Schematic diagram of the effect after point cloud classification.

[0022] Figure 5 Schematic diagram for modeling the mine area, where Figure 5 (a) is the DEM diagram of the mining area. Figure 5 (b) is a schematic diagram of the DEM profile of the mining area.

[0023] Figure 6 A schematic block diagram of the structure of a mine modeling system based on point cloud data provided by an embodiment of the present invention.

[0024] Figure 7 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the application objectives, features, and advantages of this application more obvious and easy to understand, the technical solutions protected by this application will be clearly and completely described below using specific embodiments and drawings. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this patent, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this patent.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0027] The key terms appearing in the present invention are explained below.

[0028] ICP algorithm: Iterative Closest Point, iterative closest point algorithm.

[0029] DEM: Digital Elevation Model.

[0030] Figure 1 A schematic diagram of a mine modeling method based on point cloud data provided by an embodiment of the present invention. Figure 1 The execution entity may be a point cloud data-based mine modeling system. The point cloud data-based mine modeling method provided in the embodiments of the present invention is executed by a computer device. Accordingly, the point cloud data-based mine modeling system runs on the computer device. The order of the steps in this flowchart may be changed, and some steps may be omitted, depending on different needs.

[0031] like Figure 1 As shown, the method includes the following steps.

[0032] S1, obtain the original point cloud data of the target mining area.

[0033] S2, preprocessing the original point cloud data, including voxel filtering, to obtain the target point cloud data.

[0034] S3, calculates the curvature and normal vector of each point in the target point cloud data.

[0035] S4, the coordinates, curvature, and normal vector of each point are input into the pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories.

[0036] S5, based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data, an energy function of the graph cut algorithm is constructed, and the final geological category to which each point belongs is obtained through the graph cut algorithm.

[0037] S6, construct a digital model of the target mining area based on the geological category to which each point belongs.

[0038] The mine modeling method based on point cloud data provided in this embodiment uses voxel filtering to reduce the amount of redundant point cloud data, retain key geometric structures, and improve computational efficiency. It combines curvature and normal vectors to construct a multidimensional feature description, enhancing local geometric representation capabilities (such as accurately identifying fault edges in high-curvature areas). The end-to-end classification framework based on deep learning models (such as PointNet++) integrates local details (normal vector changes) with global context (vein direction) to improve the classification accuracy of complex geological categories such as veins and fracture zones. It also outputs point-by-point probability distributions rather than hard labels, which can provide flexible constraints for post-optimization. Furthermore, a graph cut algorithm is introduced to jointly optimize classification probability and spatial smoothness through an energy function, correct isolated misclassified points (such as integrating discrete "vein" points into a continuous ore body), ensure a natural transition of geological boundaries, effectively address point cloud sparsity and noise interference, and improve classification robustness.

[0039] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process in this embodiment, another mine modeling method based on point cloud data is provided, which includes the following steps.

[0040] SS1, obtain the original point cloud data of the target mining area.

[0041] In some optional implementations, LiDAR or UAV-mounted LiDAR sensors are used, with point cloud density set to ≥50 points / ㎡ and scanning overlap rate ≥30% to ensure data integrity.

[0042] For example, a mine uses ground LiDAR to cover the entire tunnel network through multi-station scanning, and sets control points (such as reflective patches) at each station for splicing and alignment.

[0043] LiDAR parameter settings are as follows: Laser wavelength: 1,550 nm (for enhanced penetration through vegetation); Scanning frequency: 100 kHz (point cloud density ≥ 50 points / m2 per scan); Scanning angle range: horizontal ±60°, vertical ±15°.

[0044] The drone's path planning is as follows: flight altitude: 100 m (ground resolution: 0.1 m); scan overlap: 30% (adjacent routes overlap to reduce data gaps).

[0045] The acquisition device transmits the collected raw point cloud data to the backend for subsequent processing. The raw point cloud data includes coordinates (X, Y, Z), reflection intensity, and RGB color.

[0046] Figure 2 is the original point cloud and section diagram, where Figure 2 (a) is a schematic diagram of the original point cloud. Figure 2 (b) is a schematic diagram of the original point cloud section. Figure 3 Schematic diagram of DEM generated based on the original point cloud.

[0047] SS2, preprocesses the original point cloud data, including voxel filtering, to obtain the target point cloud data.

[0048] The purpose of this step is to perform preprocessing, eliminate noise and reduce data redundancy, which specifically includes the following steps.

[0049] SS2.1 performs noise reduction on the original point cloud data.

[0050] SS2.1.1, Statistical filtering: Calculate the standard deviation of local elevation and remove outliers (such as noise points).

[0051] The formula for calculating the local elevation standard deviation is as follows:

[0052] in, For the local area The elevation value of a point; is the mean elevation of the local area; is the number of points in the local area (voxel grid size: 0.5m×0.5m×0.5m).

[0053] Remove Satisfaction points (noise points).

[0054] Statistical filtering calculates the local elevation standard deviation to remove outliers (noise points), effectively identifying and eliminating interfering data. Based on the local elevation standard deviation formula, points outside the range are treated as noise points and removed, reducing the impact of noise on subsequent analysis and modeling. Point cloud data truly reflects the terrain characteristics of the target area.

[0055] SS2.1.2, Voxel filtering.

[0056] The point cloud is divided into a voxel grid of 0.1m×0.1m×0.1m, and one point is retained per voxel, reducing the data volume to 1 / 10 of the original.

[0057] Voxel filtering significantly reduces data redundancy while retaining key terrain information, reduces data storage and computing costs, and improves the efficiency of subsequent data processing and analysis without affecting the overall expression of terrain features.

[0058] SS2.2, vegetation and interference point removal.

[0059] First, the ground model is fitted using the RANSAC algorithm, and points far from the plane are marked as vegetation to achieve plane fitting classification. Then, based on the elevation threshold method, a Z value threshold is set in the dense vegetation area (such as 1m higher than the ground model) to screen vegetation points.

[0060] The RANSAC algorithm is used to fit the ground model, and vegetation points far away from the plane are marked. Then, based on the elevation threshold method, a Z value threshold is set in the dense vegetation area (such as 1m higher than the ground model) to further screen the vegetation points, so as to accurately distinguish the ground from interfering objects such as vegetation, eliminate the interference of vegetation and other objects, and improve the modeling accuracy.

[0061] SS2.3, coordinate alignment processing.

[0062] Use control points for ICP registration to eliminate the cumulative error of multi-station scanning.

[0063] Among them, the ICP iteration formula is:

[0064] Where, is the matching point in the target point cloud; is the corresponding point in the source point cloud; is the rotation matrix; is the translation vector.

[0065] Ensure that the point cloud data obtained from different scanning sites are accurately aligned in space, guarantee the overall coordinate accuracy of the point cloud data, and enable the model to accurately reflect the spatial position relationship of the target area.

[0066] SS3, calculates the curvature and normal vector of each point in the target point cloud data.

[0067] SS3.1, calculate the normal vector of each point in the target point cloud data.

[0068] In some optional implementations, the normal vector of each point in the point cloud data is calculated based on the covariance matrix method of the neighborhood points.

[0069] First, for each point in the target point cloud, neighboring points are determined using either a distance-based or point-count-based approach. The coordinates of the point's neighbors are then constructed into a matrix. The centroid coordinates of the neighbors are calculated, and the coordinates of the neighbors are centered. This is done by subtracting the centroid coordinates from each neighboring point coordinate to obtain a new coordinate matrix. The covariance matrix of the new coordinate matrix is ​​then calculated, and the eigenvalue decomposition of the covariance matrix is ​​performed to obtain the eigenvalues ​​and corresponding eigenvectors. The normal vector corresponds to the eigenvector corresponding to the minimum eigenvalue of the covariance matrix.

[0070] SS3.2, calculate the curvature of each point in the target point cloud data.

[0071] In some optional embodiments, the local curvature of the point cloud is estimated by moving least squares (MLS) , the formula is as follows.

[0072]

[0073] in, for The normal vector of the point, are three points in the local neighborhood.

[0074] In SS4, the coordinates, curvature, and normal vector of each point are input into the pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories.

[0075] The geological classification of each point is predicted using a pre-trained geological classification model. In some optional embodiments, the geological classification model employs a PointNet++ network. The network structure includes an input layer, a local set abstraction layer, MLP parameters, and a classification layer. The input layer inputs the coordinates, curvature, and normal vector of each point. The MLP parameters are [64, 64, 128], [128, 128, 256], and [256, 512]. The classification layer includes a fully connected layer (512 → 256 → 128 → number of geological categories).

[0076] When training the geological classification model, the learning rate is set to 0.001, the batch size is 32, and the loss function uses cross entropy loss.

[0077] Using point coordinates, curvature, and normal vectors as inputs comprehensively captures key features of point cloud data, including spatial location, local shape, and surface orientation. Coordinates reflect the spatial location of a point, curvature reflects the degree of local shape variation, and normal vectors indicate surface orientation. Leveraging these multi-dimensional features enables the geological classification model to learn richer and more comprehensive geological information, improving its ability to distinguish between different geological categories and resulting in more accurate predictions.

[0078] The PointNet++ network is used as the geological category classification model. The fully connected structure of the classification layer can efficiently convert the extracted features into probabilistic outputs of various geological categories rather than hard labels, which can provide flexible constraints for post-optimization.

[0079] SS5, based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, constructs the energy function of the graph cut algorithm, and obtains the final geological category of each point through the graph cut algorithm.

[0080] In some optional implementations, a graph cut algorithm is used to combine spatial adjacency and feature similarity to optimize segmentation boundaries and reduce noise interference, specifically including the following steps.

[0081] SS5.1, based on the predicted geological category of each point, the data item of the graph cut algorithm energy function is constructed, and the geological category corresponding to the maximum probability of each point belonging to various geological categories output by the geological category classification model is used as the initial predicted geological category.

[0082] SS5.2, constructs the smoothing term of the energy function of the graph cut algorithm based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data.

[0083] SS5.3, adjust the predicted geological category of each point, and obtain the final geological category of each point with the goal of minimizing the energy function.

[0084] In some optional implementations, the point cloud features of the target point cloud data include the spatial distance between two points and the normal vector of the point. Accordingly, the expression of the energy function is:

[0085] in, ; ; Where, is the smoothing coefficient, is a data item, is the smoothing term, is the total number of points in the target point cloud, for point Belongs to the geological category to which the prediction belongs The probability of is the adjacency relationship between points, is a collection of geological categories, and Points and point Belongs to the geological category The probability of for point and point The Euclidean distance between and Points and point The normal vector of is the weight coefficient, is the scale parameter.

[0086] The data item adopts the negative log-likelihood form to directly minimize the classification uncertainty. The energy function is low, driving the graph cut algorithm to adjust its labels, effectively correcting isolated misclassified points (for example, optimizing a low-confidence "vein" point to a "fault"). Inputting point-by-point multi-class probabilities rather than hard labels preserves classification confidence, providing flexible optimization space for graph cuts and avoiding the limitations of binary decision-making.

[0087] The smoothing term simultaneously integrates spatial distance and normal vector differences, and implements dual constraints through Gaussian kernel weighting, including 1) spatial continuity constraint: neighboring points tend to have similar labels, eliminating "salt and pepper noise"; 2) geometric consistency constraint: areas with large normal vector differences (such as fault edges) are allowed to have category mutations, preserving sharp geological boundaries.

[0088] Among them, the parameters Adjust the contribution weight of normal vectors to dynamically optimize smoothing strength for different geological structures (such as chaotic normal vectors in fracture zones vs. uniform normal vectors in mineral veins).

[0089] The smoothing coefficient in the above energy function Adjust the balance between data items and smoothing items to adapt to different data qualities (high noise data can increase to depend on the classification confidence). Scale parameter This approach controls the impact of neighborhoods, capturing details at a small scale (such as the microstructure of fractured zones) while smoothing macroscopic trends at a large scale (such as the extension of mineral veins). It also supports multi-label optimization algorithms such as α-expansion, solving complex segmentation problems in scenarios involving multiple interactive categories, such as mineral veins, faults, and fractured zones.

[0090] Figure 4 Schematic diagram of the effect after point cloud classification.

[0091] SS6: Build a digital model of the target mining area based on the geological category of each point. This includes the following steps.

[0092] SS6.1, divide the points belonging to the same geological category into one region to obtain multiple segmented regions.

[0093] In some optional implementations, all points are traversed and classified according to their geological category labels. A container indexed by geological category is created (e.g., a dictionary, where the key is the geological category and the value is the set of points belonging to that category). Each time a point is read, it is added to the set of points belonging to the corresponding category based on its geological category. Ultimately, each geological category corresponds to a set of points, and these sets form different segmentation regions, achieving the goal of dividing points of the same geological category into a single region.

[0094] SS6.2 uses the Poisson surface reconstruction algorithm to generate a smooth surface for each segmented area, which is the local model.

[0095] For the point cloud data in each segmented area, the Poisson surface reconstruction algorithm is applied. The algorithm first takes the point cloud data as input, uses the spatial distribution information of the point cloud, and solves the Poisson equation constructed by the point cloud density function (the solution is ). During the solution process, the algorithm analyzes the local geometric features and topological relationships of the point cloud and gradually constructs an implicit function. The zero level set of this implicit function satisfies The point set will be extracted to generate a smooth triangular mesh surface, thereby realizing the conversion from discrete point cloud to continuous smooth surface and obtaining a preliminary local model.

[0096] In some optional implementations, a parametric surface fitting operation is further performed on the portion of the segmented area that presents a regular structure. Whether the structure is regular can be determined based on the geometric characteristics of the point cloud distribution, such as whether the point cloud presents a distribution trend similar to a cylinder or a plane in space. For the portion that meets the characteristics of the regular structure, a cylindrical surface or a plane is used for fitting. Taking cylindrical surface fitting as an example, by analyzing the radial and axial distribution patterns of the point cloud, the parameters such as the radius, axis position and direction of the cylindrical surface are determined, and mathematical methods are used to fit the point cloud to the corresponding cylindrical surface. This can effectively reduce the number of polygons in the model, simplify the model structure, reduce the amount of data and computational complexity, while maintaining the model's accurate expression of the actual geological structure, making the local model more concise and efficient.

[0097] In SS6.3, the ICP algorithm is used to merge the local models into an overall model in a unified coordinate system.

[0098] In some optional embodiments, one of the multiple local models is selected as a reference model, and its coordinate system is determined to be the benchmark of the unified coordinate system. For each of the remaining local models, the iterative closest point (ICP) algorithm is used for alignment. First, the nearest neighbor points are found for the points of the current local model in the reference model; then, based on these corresponding point pairs, the optimal rotation matrix R and translation vector t are calculated by minimizing error functions such as the mean square error, so that the current local model is closer to the reference model after transformation; this process of finding the nearest neighbor points and calculating the transformation is repeated until the convergence conditions are met (such as the transformation error is less than a preset threshold or the maximum number of iterations is reached). After performing the above operations on all local models in sequence, they are merged into an overall model under the unified coordinate system.

[0099] SS6.4, uses the moving least squares interpolation method to fill in the data missing areas of the overall model to obtain a digital model of the target mining area.

[0100] First, the entire model is traversed, identifying areas with missing data by determining the distribution of points or setting specific conditions (such as areas where point density falls below a certain threshold). For each point to be interpolated within the missing area, a local neighborhood is defined with it as the center (the radius can be determined based on distance). A weight function is constructed within the neighborhood. The weight function is generally related to the distance from the point to be interpolated, with a greater weight being assigned to closer distances, such as the commonly used Gaussian weight function.

[0101] Within the local neighborhood, the coordinates of the interpolated points are calculated using a weighted least squares fit, based on the principle of moving least squares. This process is repeated for all interpolated points within the missing region, completing the missing areas in the overall model and ultimately obtaining a digital model of the target mining area.

[0102] In some optional embodiments, after the digital model of the target mining area is constructed in step SS6, automated and interactive corrections are performed. Automated corrections include topology checks to detect holes and cracks in the model and automatically repair or mark them. They also include geological attribute annotations, adding attributes such as coal seam thickness and fault dip based on feature extraction results. Interactive corrections include adjusting local surface smoothness or geological boundaries using 3D modeling software (such as CloudCompare). Figure 5 Schematic diagram for modeling the mine area, where Figure 5 (a) is the DEM diagram of the mining area. Figure 5 (b) is a schematic diagram of the DEM profile of the mining area.

[0103] For an open-pit mine (2 km² area, 120 million point cloud points), the following results were obtained through experimental data comparison.

[0104]

[0105] This method requires only nine hours, significantly reducing processing time. This is due to the use of more efficient algorithms and strategies throughout the process. For example, in the preprocessing step, voxel filtering and other methods are used to quickly remove noise and redundant data, reducing subsequent computational effort. During model construction, optimized algorithms and data structures are employed to accelerate point cloud analysis and model generation, resulting in an 80.25% reduction in processing time compared to traditional methods. By comprehensively utilizing multi-dimensional features of point cloud data, such as coordinates, curvature, and normal vectors, and employing advanced geological classification models (such as the PointNet++ network), this method can more comprehensively and accurately learn and identify fault characteristics, effectively reducing false positives and missed detections, and achieving a 27.9% improvement in accuracy compared to traditional methods. Furthermore, through the coordinated integration of preprocessing, feature calculation, and model construction, the model more accurately reflects the actual mine topography, achieving a 66.7% improvement in accuracy compared to traditional methods.

[0106] An embodiment of a mine modeling method based on point cloud data is described in detail above. Based on the mine modeling method based on point cloud data described in the above embodiment, an embodiment of the present invention also provides a mine modeling device based on point cloud data corresponding to the method.

[0107] Figure 6 This is a block diagram illustrating the structure of a point cloud data-based mine modeling system according to an embodiment of the present invention. In this embodiment, the point cloud data-based mine modeling system 600 can be divided into multiple functional modules according to their functions. A module, as used herein, refers to a series of computer program segments that can be executed by at least one processor and perform a fixed function, and is stored in a memory.

[0108] The original point cloud acquisition module 610 is used to acquire original point cloud data of the target mining area.

[0109] The point cloud pre-processing module 620 is used to pre-process the original point cloud data, including voxel filtering, to obtain target point cloud data.

[0110] The point cloud feature calculation module 630 is used to calculate the curvature and normal vector of each point in the target point cloud data.

[0111] The geological category prediction module 640 is used to input the coordinates, curvature, and normal vector of each point as input values ​​into a pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories.

[0112] The geological category determination module 650 is used to construct an energy function of a graph cut algorithm based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, and obtain the final geological category of each point through the graph cut algorithm.

[0113] The digital model building module 660 is used to build a digital model of the target mining area based on the geological category to which each point belongs.

[0114] The mine modeling system based on point cloud data of this embodiment is used to implement the aforementioned mine modeling method based on point cloud data. Therefore, the specific implementation method of this system can be seen in the embodiment part of the mine modeling method based on point cloud data in the previous text. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part, and will not be elaborated here.

[0115] In addition, since the mine modeling system based on point cloud data of this embodiment is used to implement the aforementioned mine modeling method based on point cloud data, its function corresponds to that of the aforementioned method and will not be described in detail here.

[0116] Figure 7 A schematic diagram of the structure of a terminal 700 provided in an embodiment of the present invention includes: a processor 710, a memory 720, and a communication unit 730. The processor 710 is configured to implement the following steps when executing a mine modeling program based on point cloud data stored in the memory 720: Obtain original point cloud data of the target mining area; Preprocess the original point cloud data, including voxel filtering, to obtain the target point cloud data; Calculate the curvature and normal vector of each point in the target point cloud data; The coordinates, curvature, and normal vector of each point are input into a pre-trained geological classification model to obtain the probability that each point belongs to various geological categories; Based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, the energy function of the graph cut algorithm is constructed, and the final geological category of each point is obtained through the graph cut algorithm; A digital model of the target mining area is constructed based on the geological category to which each point belongs.

[0117] The terminal 700 includes a processor 710, a memory 720, and a communication unit 730. These components communicate via one or more buses. Those skilled in the art will appreciate that the server structure shown in the figure does not limit the present invention; it may be a bus structure or a star structure, and may include more or fewer components than shown, or combine certain components, or arrange the components differently.

[0118] The memory 720 can be used to store execution instructions of the processor 710. The memory 720 can be implemented by any type of volatile or non-volatile storage terminal, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 720 are executed by the processor 710, the terminal 700 can perform some or all of the steps in the following method embodiments.

[0119] The processor 710 is the control center of the storage terminal. It uses various interfaces and lines to connect various parts of the entire electronic terminal. It executes various functions of the electronic terminal and / or processes data by running or executing software programs and / or modules stored in the memory 720, and calling data stored in the memory. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of multiple packaged ICs with the same or different functions. For example, the processor 710 can only include a central processing unit (CPU). In the embodiment of the present invention, the CPU can be a single computing core or multiple computing cores.

[0120] The communication unit 730 is configured to establish a communication channel so that the storage terminal can communicate with other terminals, receive user data sent by other terminals, or send user data to other terminals.

[0121] The present invention also provides a computer storage medium, wherein the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0122] The computer storage medium stores a mine modeling program based on point cloud data. When the mine modeling program based on point cloud data is executed by a processor, the following steps are implemented: Obtain original point cloud data of the target mining area; Preprocess the original point cloud data, including voxel filtering, to obtain the target point cloud data; Calculate the curvature and normal vector of each point in the target point cloud data; The coordinates, curvature, and normal vector of each point are input into a pre-trained geological classification model to obtain the probability that each point belongs to various geological categories; Based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, the energy function of the graph cut algorithm is constructed, and the final geological category of each point is obtained through the graph cut algorithm; A digital model of the target mining area is constructed based on the geological category to which each point belongs.

[0123] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software and a necessary general-purpose hardware platform. Based on this understanding, the technical solutions in the embodiments of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, among other media capable of storing program code, and includes instructions for causing a computer terminal (which can be a personal computer, a server, or a second terminal, a network terminal, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention.

[0124] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0125] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0126] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0127] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A mine modeling method based on point cloud data, characterized in that: The following steps are involved: Obtain original point cloud data of the target mining area; Preprocess the original point cloud data, including voxel filtering, to obtain the target point cloud data; Calculate the curvature and normal vector of each point in the target point cloud data; The coordinates, curvature, and normal vector of each point are input into a pre-trained geological classification model to obtain the probability that each point belongs to various geological categories; Based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, the energy function of the graph cut algorithm is constructed, and the final geological category of each point is obtained through the graph cut algorithm; A digital model of the target mining area is constructed based on the geological category to which each point belongs.

2. The mine modeling method based on point cloud data according to claim 1, characterized in that: The energy function of the graph cut algorithm is constructed based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data. The final geological category of each point is obtained through the graph cut algorithm, which includes: Based on the predicted geological category of each point, the data item of the energy function of the graph cut algorithm is constructed, and the geological category corresponding to the maximum probability of each point belonging to various geological categories output by the geological category classification model is used as the initial predicted geological category; The smoothing term of the energy function of the graph cut algorithm is constructed based on the probability that each point belongs to various geological categories and the point cloud features of the target point cloud data; The predicted geological category of each point is adjusted, and the final geological category of each point is obtained with the goal of minimizing the energy function.

3. The mine modeling method based on point cloud data according to claim 2, characterized in that: The point cloud features of the target point cloud data include the spatial distance between two points and the normal vector of the point.

4. The mine modeling method based on point cloud data according to claim 3, characterized in that: The expression of the energy function is: in, ; ； Where, is the smoothing coefficient, is a data item, is the smoothing term, is the total number of points in the target point cloud, for point Belongs to the geological category to which the prediction belongs The probability of is the adjacency relationship between points, is a collection of geological categories, and Points and point Belongs to the geological category The probability of for point and point The Euclidean distance between and Points and point The normal vector of is the weight coefficient, is the scale parameter.

5. The mine modeling method based on point cloud data according to any one of claims 1 to 4, characterized in that: The geological category classification model is constructed by the PointNet++ network.

6. The mine modeling method based on point cloud data according to any one of claims 1 to 4, characterized in that: Construct a digital model of the target mining area based on the geological category of each point, including: Divide the points belonging to the same geological category into one region to obtain multiple segmented regions; A Poisson surface reconstruction algorithm is used to generate a smooth surface for each segmented area, which is the local model. The ICP algorithm is used to merge the local models into an overall model in a unified coordinate system; The moving least squares interpolation method is used to fill the data missing areas of the overall model to obtain a digital model of the target mining area.

7. The mine modeling method based on point cloud data according to any one of claims 1 to 4, characterized in that: The preprocessing of the original point cloud data also includes statistical filtering, removing vegetation and interference points through plane fitting classification and elevation threshold method, and using control points for ICP algorithm alignment.

8. A mine modeling system based on point cloud data, characterized in that: include: The original point cloud acquisition module is used to obtain the original point cloud data of the target mining area; Point cloud preprocessing module, used to preprocess the original point cloud data, including voxel filtering, to obtain target point cloud data; Point cloud feature calculation module, used to calculate the curvature and normal vector of each point in the target point cloud data; The geological category prediction module is used to input the coordinates, curvature, and normal vector of each point into the pre-trained geological category classification model to obtain the probability that each point belongs to various geological categories; The geological category determination module is used to construct the energy function of the graph cut algorithm based on the probability of each point belonging to various geological categories and the point cloud features of the target point cloud data, and finally obtain the geological category to which each point belongs through the graph cut algorithm; The digital model building module is used to build a digital model of the target mining area based on the geological category to which each point belongs.

9. A terminal, characterized in that: include: A memory for storing a mine modeling program based on point cloud data; A processor is configured to implement the steps of the mine modeling method based on point cloud data as claimed in any one of claims 1 to 7 when executing the mine modeling program based on point cloud data.

10. A computer-readable storage medium, characterized in that The readable storage medium stores a mine modeling program based on point cloud data. When the mine modeling program based on point cloud data is executed by a processor, the steps of the mine modeling method based on point cloud data according to any one of claims 1 to 7 are implemented.

Images