A ground point bottom model establishing method based on a point cloud minimum elevation clustering algorithm

Abstract

The application provides a ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm, which is divided into the following steps: open pit point cloud data collection, open pit point cloud data preprocessing, open pit point cloud data gridding, open pit point cloud bottom model establishment based on a DBSCAN density clustering algorithm, missing point cloud interpolation, and open pit final ground point bottom model point cloud data establishment. Specifically, first, open pit full point cloud data is collected, and the point cloud data is subjected to thinning processing; second, the point cloud data is gridded, all grid point cloud data is traversed, and the DBSCAN density clustering number of the lowest point in the grid is calculated in the order of elevation from low to high, and when the clustering number reaches N, the point is considered to be the minimum ground point; finally, missing point cloud is filled in through a nearest neighbor interpolation method, and then the final open pit ground bottom model point cloud data is obtained. The method aims to calculate the DBSCAN density clustering number of the lowest point in the grid, so that the open pit ground bottom model point cloud data can be quickly and accurately collected, which provides important guidance value for the establishment of an open pit DEM model by using an auxiliary progressive morphological filter and the intelligent acceptance of open pit blasting.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of geographic information systems and remote sensing technology and application, in particular to the rapid and accurate establishment of a ground point bottom model of an open pit, and more particularly to a ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm. BACKGROUND

[0002] In the calculation process of stripping volume of an open pit, the conventional calculation method usually relies on manual measurement or intermittent data collection, which has problems such as discontinuous data collection, low calculation accuracy, long time consumption, and large subjective errors, and is difficult to meet the needs of modern open pits for high precision, real-time and automated management.

[0003] With the rapid development of geographic information technology and remote sensing disciplines, point cloud technology, as a new means, is gradually applied to the accurate calculation of stripping volume of an open pit. Point cloud technology can provide more accurate basic data for stripping volume calculation by obtaining three-dimensional spatial data of the open pit through high-precision laser scanning. However, the use of point cloud technology to calculate stripping volume usually requires differential operation based on pre-blasting and post-blasting digital elevation models (DEMs) to obtain volume change information. In the process of establishing a DEM model, the extraction of the minimum bottom model is the key first step, and its accuracy directly affects the reliability of the subsequent DEM model and the accuracy of the stripping volume calculation. The extraction of the minimum bottom model aims to filter out the lowest ground points from complex point cloud data, excluding elevation outliers and noise interference, thereby providing high-quality basic data for the construction of the DEM model. Therefore, it is of great significance to study an efficient and accurate minimum bottom model extraction method for improving the application value of point cloud technology in stripping volume calculation.

[0004] Based on the above problems, there is an urgent need for a ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm to effectively meet the needs of modern open pits for high precision, real-time and automated management. This method has important theoretical and engineering value for efficient acceptance and information-based green mine construction of mine stripping. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application proposes a ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm to solve the engineering problems described in the background.

[0006] The present application accurately extracts the lowest ground points by efficient point cloud data processing and density clustering technology, and excludes elevation outliers and noise interference. Specifically, the following technical solutions are implemented.

[0007] A ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm, characterized in that the method comprises the following steps:

[0008] S1: stope point cloud data collection; the unmanned aerial vehicle is equipped with a photogrammetry lens, and parameters such as a stope range, a take-off height, a take-off speed, a flight path, and a lateral overlap rate are set, so as to obtain original point cloud data of the stope;

[0009] S2: data preprocessing; the original point cloud data is processed by the method of equal-interval thinning;

[0010] S3: stope point cloud data gridding; the full-stope point cloud is gridded, and the gridding horizontal and vertical dimensions are set as the same fixed value;

[0011] In S3, the gridding size of the stope point cloud data is equal to the thinning interval, and only the X and Y plane point clouds need to be gridded, and the coordinates of any gridded point cloud are X i , Y i as follows:

[0012]

[0013] In the formula, (x, y) represents the plane coordinates of the point, X i , Y i represents the gridded coordinates, and voxel_ size represents the gridding size;

[0014] S4: stope point cloud bottom model establishment based on the DBSCAN density clustering algorithm; all gridded point cloud data are traversed, and the DBSCAN density clustering number of the lowest point in the grid is calculated in the order of elevation from low to high, and when the total number of core points and boundary points in the clustering result is less than the preset point cloud number N, it means that the lowest point and the cluster where the lowest point is located are noise points, and when the total number of core points and boundary points in the clustering result meets the preset point cloud number N, it means that the lowest point is a ground point bottom model point in the grid;

[0015] Further, in S4, the DBSCAN density clustering algorithm is applied to the point cloud in each grid, and the specific steps are as follows:

[0016] S401: initialization parameters; ε represents the neighborhood radius, and Q represents the minimum number of points required in the neighborhood;

[0017] S402: DBSCAN density clustering is performed on the lowest point in the grid in the order of elevation from low to high, and all gridded points are traversed and clustered;

[0018] Further, the DBSCAN density clustering method in S402 is as follows:

[0019] S4021: the points in the grid are marked as P i in the order of elevation from low to high, i=1, 2, 3, 4…n, and the point P iThe ε-neighborhood of, that is, all points whose distance from point P i is less than or equal to ε;

[0020] S4022: Expand P i , i = 1, 2, 3, 4…n, clusters; The number of points within the ε-neighborhood of point P i , i = 1, 2, 3, 4…n, being ≥ Q is denoted as a core point, the number of points within the ε-neighborhood being < Q and the point being within the ε-neighborhood of a certain core point indicates that this P point is a boundary point, and the number of points within the ε-neighborhood of point P being < Q and the point not being within the ε-neighborhood of any core point indicates that this point is a noise point;

[0021] S4023: When P i , i = 1, 2, 3, 4…n, is expanded into a cluster and the scale of the cluster is greater than the specified quantity N, it can be determined that point P i is the minimum ground point. When P i , i = 1, 2, 3, 4…n, is not expanded into a cluster, or the scale of the formed cluster is not greater than the specified quantity N, it can be determined that point P i is a noise point;

[0022] S403: Determine the minimum ground point according to the neighborhood characteristics of the points within the grid;

[0023] S5: Missing point cloud interpolation; Fill the missing point cloud within the grid by the method of nearest neighbor interpolation;

[0024] S6: Establishment of the point cloud data of the final ground bottom model of the stope.

[0025] Furthermore, the nearest neighbor interpolation method for compensating missing points needs to search for the topological information characteristics of the nearest point cloud of the missing point cloud through a kd-tree, and then solve the information of the missing point cloud according to the topological information characteristics of the nearest point cloud.

[0026] In the said S4023, when P i , i = 1, 2, 3, 4…n, is not expanded into a cluster, take the point with the lowest elevation outside P i , i = 1, 2, 3, 4…n, within the grid range and its extended scattered points to continue to expand the cluster once, thereby reducing the running calculation amount.

[0027] The present invention provides a method for establishing a ground point bottom model based on a point cloud minimum elevation clustering algorithm, having the following technical effects:

[0028] 1. By sequentially calculating the DBSCAN density clustering quantity of the lowest points within the grid, rapid and accurate acquisition of the point cloud data of the stope ground bottom model is achieved, providing important guiding value for assisting in the establishment of the stope DEM model by progressive morphological filtering;

[0029] 2. The ground point bottom model extraction process can efficiently remove noise points and noise clusters by the point cloud minimum elevation clustering algorithm, and only needs to set the noise threshold according to the point cloud density, without specifying the number of noise clusters;

[0030] 3. It provides an important reference for establishing a similar open-pit mine digital elevation model (DEM), and has important theoretical and engineering significance for green mine construction and intelligent acceptance of stripping volume. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0032] Figure 1 The flow chart of the ground point bottom model establishment method based on the point cloud minimum elevation clustering algorithm is shown in the present application.

[0033] Figure 2 The bottom model noise point distribution diagram of the original point cloud data of the open-pit mine is shown.

[0034] Figure 3 The schematic diagram of discrete noise points and noise clusters is shown.

[0035] Figure 4 The color scale schematic diagram of the pit bottom model is shown. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0037] The purpose of the present application is to provide a ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm.

[0038] Embodiment one:

[0039] Taking a three-dimensional original point cloud model of a certain open-pit mine as an example.

[0040] S1: Pit point cloud data acquisition. After the unmanned aerial vehicle is equipped with a photogrammetry lens, the pit range is circled, the take-off height is 200m, the ground flight height is 200m, the take-off speed is 12m / s, the lateral overlap rate is 80%, and the heading overlap rate is 70%, to obtain the original point cloud data of the pit;

[0041] S2: data preprocessing. The original point cloud data is thinned by an equal interval thinning method, and the equal interval interval is set to 20 cm;

[0042] S3: stope point cloud data gridding. The full stope point cloud is gridded, and the gridding horizontal and vertical dimensions are set to the same value, specifically 20 cm;

[0043] S4: stope point cloud bottom model establishment based on DBSCAN density clustering algorithm; all grid point cloud data is traversed, and the DBSCAN density clustering number of the lowest point in the grid is calculated in turn according to the elevation from low to high in the grid. When the sum of core points and boundary points in the clustering result is less than the preset point cloud number N, it means that the lowest point and the cluster where the lowest point is located are noise points. When the sum of core points and boundary points in the clustering result meets the preset point cloud number N, it means that the lowest point is the ground point bottom model point in the grid;

[0044] S5: missing point cloud interpolation. The missing point cloud in the grid is filled by the nearest neighbor interpolation method;

[0045] S6: stope final ground bottom model point cloud data establishment.

[0046] Data preprocessing mainly reflects the equal interval thinning of stope point cloud data. The point cloud interval after thinning is directly related to the point cloud density in the lowest point DBSCAN density clustering in S4, and is uniformly set to 20 cm in embodiment one.

[0047] The grid size in stope point cloud data gridding is equal to the thinning interval, and only the X, Y plane point cloud needs to be gridded. The coordinates of any point cloud data after gridding are X i , Y i as follows:

[0048]

[0049] Where (x, y) represents the plane coordinates of the point, X i , Y i represents the gridded coordinates, voxel_ size represents the gridding size.

[0050] The DBSCAN density clustering algorithm is applied to each grid point cloud, and the specific steps are as follows:

[0051] S401: initialization parameters; ε represents the neighborhood radius, which is set to a sphere with a radius of 40 cm (the radius needs to be greater than the grid size), and Q represents the minimum number of points required in the neighborhood, which is set to 10;

[0052] S402: Perform DBSCAN density clustering on the lowest point in the grid in order of elevation from low to high, and traverse all grid points to perform clustering;

[0053] S403: Determine the smallest ground point based on the neighborhood characteristics of points within the grid.

[0054] The DBSCAN density clustering method in S402 is as follows:

[0055] S4021: Within a grid, points are marked as P in ascending order of elevation. i Let i = 1, 2, 3, 4…n, calculate point P. i The ε-neighborhood of point P, i.e. i All points whose distance is less than or equal to ε;

[0056] S4022: Extended P i i = 1, 2, 3, 4…n, cluster; point P i For i = 1, 2, 3, 4...n, a point with ≥ 10 points in its ε-neighborhood is considered a core point. A point with < 10 points in its ε-neighborhood and that is within the ε-neighborhood of a core point is considered a boundary point. A point P with < 10 points in its ε-neighborhood and that is not within the ε-neighborhood of any core point is considered a noise point.

[0057] S4023: When P i When i = 1, 2, 3, 4…n, and the cluster is expanded to a number N, point P can be determined if the cluster size is greater than the specified number N. i For the minimum ground point, when P i If i = 1, 2, 3, 4…n, and the cluster does not expand into a cluster, or the size of the cluster does not exceed the specified number N, then point P can be determined. i For noise, P is taken within the grid area. i The points i = 1, 2, 3, 4…n, and the points with the lowest elevations outside the expanded scatter plots, are then expanded into a cluster, thereby reducing the computational load.

[0058] In this first embodiment, N is set to 100.

[0059] The specific principle is as follows: Figure 3 The schematic diagram of discrete noise points and noise clusters is shown below:

[0060] To more clearly illustrate the DBSCAN density clustering process, the minimum number of points Q to be reached within the neighborhood in the diagram is 4.

[0061] First, a point cloud cluster is established; point A (circle) represents the core point, and there are at least 4 point cloud data within the neighborhood radius ε. If the number of points in the neighborhood is less than 4 and the point is within the ε neighborhood of a certain core point, it is considered that the point is a boundary point (triangle) of point C. The core point and the boundary point form a cluster.

[0062] Then, it is judged whether the cluster is a noise cluster; the number of point clouds constituting the cluster is calculated, when the size of the cluster is < N (taking 12), it is a noise cluster, when the size of the cluster is ≥ N (taking 12), the cluster is represented as a cluster containing all points of the ground points.

[0063] Figure 3 The bottom cluster in the schematic diagram of discrete noise points and noise clusters has a size of 10 < N (taking 12), and is therefore judged as a noise cluster.

[0064] Finally, it is judged again along the elevation direction whether the point with the minimum elevation constitutes a cluster and whether the size of the cluster satisfies ≥ N (taking 12), when it satisfies, it is judged that the point with the minimum elevation is a ground point, and the N points are judged as discrete noise points because they do not satisfy the number of 4 point clouds that can be reached in the neighborhood radius ε for constituting a cluster.

[0065] A ground point bottom model establishment method based on a point cloud minimum elevation clustering algorithm, mainly according to the principle that:

[0066] 1. The ground point point cloud presents as a core point and a boundary point in the DBSCAN density clustering, and the size of the cluster is large;

[0067] 2. Discrete noise points cannot be ground points;

[0068] 3. The size of a noise cluster is small.

[0069] The nearest neighbor interpolation method compensates for missing points by searching the nearest point cloud topological information features of the missing point cloud through the kd-tree, and then solving the missing point cloud information according to the nearest point cloud topological information features.

[0070] In the specification, each embodiment is described in a progressive manner, and each embodiment mainly explains the difference from other embodiments, and the same or similar parts between each embodiment can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the related parts can be referred to the method part.

[0071] In this paper, specific examples are applied to explain the principles and implementation modes of the present application, and the above embodiment description is only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In summary, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for establishing a ground point base model based on a point cloud minimum elevation clustering algorithm, characterized in that, The method comprises the following steps: S1: collecting stope point cloud data; setting the stope range, take-off height, take-off speed, flight path, and lateral overlap rate after the unmanned aerial vehicle is equipped with a photogrammetry lens to obtain original stope point cloud data; S2: data preprocessing; the original point cloud data is processed by the equal-interval thinning method; S3: stope point cloud data gridding; the full-stope point cloud is gridded, and the horizontal and vertical dimensions of the grid are set to the same fixed value; In the S3 stope point cloud data rasterization, the grid size and the thinning interval are equidistant, and only the X and Y plane point clouds need to be rasterized. The coordinates of any point cloud data after rasterization are X i , Y i : where (x, y) represents the planar coordinates of the point, X i , Y i represents the rasterized coordinates, voxel size represents the rasterized size; S4: establishing a stope point cloud bottom model based on the DBSCAN density clustering algorithm; all grid point cloud data is traversed, and the DBSCAN density clustering number of the lowest point in the grid is calculated in the order of elevation from low to high; when the total number of core points and boundary points in the clustering result is less than the preset point cloud number N, it means that the lowest point and the cluster where the lowest point is located are noise points; when the total number of core points and boundary points in the clustering result meets the preset point cloud number N, it means that the lowest point is a ground point bottom model point in the grid; In S4, the DBSCAN density clustering algorithm is applied to the point cloud in each grid, and the specific steps are as follows: S401: initializing parameters; ε represents the neighborhood radius, and Q represents the minimum number of points required in the neighborhood; S402: DBSCAN density clustering is performed on the lowest point in the grid in the order of elevation from low to high, and all grid points are traversed for clustering; In S402, the DBSCAN density clustering method is as follows: S4021: mark as point P in the grid in order from low to high elevation i , i = 1, 2, 3, 4…n, calculate the ε neighborhood of point P i , that is, all points with a distance less than or equal to ε from point P i ; S4022: Extend P i , i = 1, 2, 3, 4…n, cluster; point P i , i = 1, 2, 3, 4…n, the number of points in the ε neighborhood of P is ≥ Q, which is expressed as a core point, the number of points in the ε neighborhood of P is < Q and the point is in the ε neighborhood of a core point, which means that the point P is a boundary point, the number of points in the ε neighborhood of P is < Q and the point is not in the ε neighborhood of any core point, which means that the point is a noise point; S4023: When P i , i = 1, 2, 3, 4…n, is extended to a cluster, and the size of the cluster is greater than a specified number N, it can be judged that point P i is the minimum ground point; when P i , i = 1, 2, 3, 4…n, is not extended to a cluster, or the size of the cluster is not greater than the specified number N, it can be judged that point P i is a noise point; S403: determining the minimum ground point according to the neighborhood characteristics of the points in the grid; S5: missing point cloud interpolation; the missing point cloud in the grid is filled by the nearest neighbor interpolation method; S6: establishing a final ground bottom model point cloud data of the stope.

2. The ground point bottom model establishment method based on the point cloud minimum elevation clustering algorithm according to claim 1, wherein the data preprocessing in S2 mainly reflects equal-interval thinning of the stope point cloud data, and the interval of the thinned point cloud is directly related to the point cloud density in the minimum point DBSCAN density clustering in S4.

3. The ground point bottom model establishment method based on the point cloud minimum elevation clustering algorithm according to claim 1, wherein the nearest neighbor interpolation method compensates for the missing points by searching the nearest point cloud topological information features of the missing point cloud through the kd-tree, and then solving the missing point cloud information according to the nearest point cloud topological information features.

4. The method for establishing a ground point-base model based on the point cloud minimum elevation clustering algorithm according to claim 1, wherein S4023 when P i If i = 1, 2, 3, 4…n, and there is no expansion into a cluster, then P is taken within the raster area. i The points i = 1, 2, 3, 4...n, and the points with the lowest elevations outside the scatter plots are expanded into a cluster, thereby reducing the amount of computation.

Images