A method for obtaining the block size of blasted ore

Abstract

The present invention discloses a method for obtaining the block size of blasted ore, belonging to the technical field of mine mining. The method for obtaining the block size of blasted ore includes the following steps: S1. Segmenting cloud data based on the VCCS algorithm improved by point cloud curvature: Establishing a voxelized grid, substituting the point cloud curvature into the supervoxel segmentation process, and using the VCCS algorithm to segment the cloud data of blasted points from a three-dimensional level to obtain the first segmented data; S2. Strengthening the ore boundary information based on the edge extraction algorithm and then using the CPC algorithm to cluster the first data to obtain the second data; S3. Calculating the block size of blasted ore: Blocking and outputting the second data of each piece of ore and calculating the point cloud diameter, and obtaining the size of the ore block size by calculating the three-dimensional space distance between the two farthest points in the second data of each clustering result. This method can accurately obtain the size of the blasted block size.

Description

Technical Field

[0001] The invention relates to the technical field of mining, and in particular to a method for obtaining the block size of blast pile ore. Background Art

[0002] Mineral resources occupy a very important position in the national economy. The supply of mineral resources is closely related to the country's resource security, economic security, and military security. In recent years, my country's development and utilization of mineral resources has continued to develop, mining technology and methods have been continuously improved, and production has become more standardized and scientific. Open-pit mining plays a pivotal role in the entire mineral resource mining in my country. For open-pit mines, they have the characteristics of strong production capacity, large production scale, high degree of mechanization, and high degree of large-scale equipment. Due to the large scale of open-pit mine production, blasting methods are often used to collapse and crush the ore rock for easy mining. After the mine blasting, a blast pile will be formed, and the blast pile block size is one of the important indicators to measure the blasting effect. Too large a block size will increase the number of secondary blasting, increase mining costs, and affect production efficiency. Too small a block size will increase the amount of waste slag and reduce economic income. Therefore, by calculating the blast pile block size to feedback the blasting effect, so as to adjust the blasting parameters so that the blast pile block size reaches a suitable range, it is of great significance for the production and development of mines.

[0003] Nowadays, many open-pit mines identify explosive piles based on two-dimensional imaging systems, but this method still has many shortcomings: first, it requires tedious manual editing. The automatic edge delineation tool of the two-dimensional imaging system needs to clearly grasp the edge of each ore, but its effect is not ideal, so manual editing is necessary; second, it is difficult for the two-dimensional imaging system to obtain the entire explosive pile block size distribution. The coverage range of a two-dimensional image system is limited, and multiple photos are required to obtain a more convincing explosive pile block size distribution; finally, the two-dimensional image system cannot identify overlapping ores, and it can only be analyzed on a plane, resulting in a complex calculation process for the explosive pile block size and inaccurate calculation results. Summary of the invention

[0004] The purpose of the present invention is to overcome the above technical deficiencies, provide a method for obtaining the block size of explosive pile ore, and solve the technical problem that the block size of explosive pile ore cannot be accurately obtained in the prior art.

[0005] In order to achieve the above technical purpose, the technical solution of the present invention provides a method for obtaining the block size of blast pile ore, comprising the following steps:

[0006] S1. VCCS algorithm for segmenting cloud data improved based on point cloud curvature: Establish a voxelized grid, substitute the point cloud curvature into the supervoxel segmentation process, set the weight value parameter set before supervoxel segmentation as the curvature weight value, and then use the VCCS algorithm that performs voxelization processing on point cloud data based on octree and uses the K-means clustering algorithm for segmentation to segment the cloud data of the muck pile points from the three-dimensional level to obtain the first segmented data;

[0007] S2. Strengthen the ore boundary information based on the edge extraction algorithm and then use the CPC algorithm to perform clustering processing on the first data to obtain the second data; The input data of this algorithm is based on the first data segmented by the VCCS algorithm, and the second data is obtained by recursively segmenting through the concavity or convexity between two adjacent supervoxels;

[0008] S3. Calculate the block size of the muck pile ore: Output the second data of each piece of ore and calculate the point cloud diameter. The size of the ore block size is obtained by calculating the three-dimensional space distance between the two farthest points in the second data of each clustering result.

[0009] Further, in some embodiments, in step S1, the cloud data is obtained by the following steps:

[0010] Use a three-dimensional laser scanner to perform multi-station scanning and acquisition of the cloud data before and after mine blasting, and process the acquired cloud data with RiSCAN PRO or CloudCompare software.

[0011] Further, in some embodiments, in step S3, the three-dimensional space distance between the two farthest points is obtained by the following steps:

[0012] S31. Randomly select a point in the ore as the starting point A, select other adjacent points as the ending point B, calculate the distance between point A and point B, denoted as dis1, and record the starting point A, the ending point B, and the current distance Dis at this time. At this time, Dis = dis1;

[0013] S32. Replace any one of the two points A and B with an adjacent point C of these two points, and then calculate the distance between the un-replaced point A or point B and point C, denoted as dis2;

[0014] S33. Compare dis1 and dis2; if dis1 is greater than dis2, then keep the current point pair and distance, that is, the recorded point pair is still the starting point A, the ending point B and the distance dis1 at this time; if dis2 is greater than dis1, then the recorded point pair and distance should be updated in time, and Dis = dis2 at this time; repeat the above steps until all points in the ore point cloud are traversed. As the traversal process continues, the calculated distance is gradually converging to the three-dimensional space distance between the two farthest points.

[0015] Further, in some embodiments, in step S2, the strengthening of the ore boundary information based on the edge extraction algorithm includes: first, calculate the surface normal of each point in the muck pile points, make the tangent plane of this point according to the point to be calculated and its normal vector, then construct a local coordinate system with this point as the coordinate center, calculate the angle between the vector from other points to this point and the established coordinate axes in the clockwise direction, compare the maximum value of the difference between adjacent angles and the angle threshold. If the maximum value of the difference is greater than the angle threshold, then this point is an edge point. Repeat the above steps to calculate each point in the muck pile, and finally realize edge extraction.

[0016] Further, in some embodiments, in step S1, the point cloud curvature is obtained by the following steps:

[0017] Local fitting of normal curvature: Assume that any point m in the muck pile point cloud has n neighboring points, and the normal vector of the i-th neighboring point Pi is N i , taking point m as the origin, and taking the orthogonal unit vectors X, Y of the coordinates of point m and its unit normal vector M as the coordinate axes, establish the local coordinate system m-XYM of m. In this coordinate system, the coordinates of the neighboring point P i are (x i , y i , z i ), and the coordinates of the normal vector N i of the neighboring point are (n x，i , n y，i , n z，i ). Then, the normal curvature of point m can be calculated through the osculating circle passing through point m as

[0018] According to Euler's formula, obtain the relationship between normal curvature and principal curvature:

[0019]

[0020] In the above formula, is the angle between the tangent of the normal section of point m passing through point p i and the principal direction; q1 and q2 are the two principal curvatures of point m.

[0021] Further, in some embodiments, point m relative to the neighboring point pi The formula for normal curvature is as follows:

[0022]

[0023] Where the α angle is the angle between the normal vector M of point m and mp i and the β angle is the angle between the normal vector M of point m and the normal vector N of the neighboring point i The coordinates of the neighboring point P i are (x i , y i , z i ), and the coordinates of the normal vector N of the neighboring point i are (n x，i , n y，i , n z，i ), corresponding to x i , y i , n y,i , n x,i in the formula respectively.

[0024] Furthermore, in some embodiments, before obtaining the relationship between the normal curvature and the principal curvature according to Euler's formula, it further includes: least squares fitting of Euler's equation:

[0025] Assume that the XY coordinates and the normal vector M at the given point m are respectively:

[0026]

[0027] Assume further that point m is on plane L with its normal vector being M. Let e1 and e2 be the principal directions at point m, and the corresponding principal curvatures be k1 and k2. Let the parameter be the angle between vectors e1 and e2, and i be the angle between vector X and vector mP.

[0028] Furthermore, in some embodiments, in step S1, before adopting the VCCS algorithm, it further includes establishing a voxelized grid with the following parameter settings: voxel resolution Rvoxel = 0.025, grid fixed resolution Rseed = 0.1, distance weight Ws = 0.4, normal vector weight Wn = 0.3. After the grid is set up, run the VCCS algorithm to start segmentation.

[0029] Furthermore, in some embodiments, in step S2, the maximum number of cuts of the CPC algorithm is 1000, the minimum size is set to 0.1, the minimum score that the cut must reach is set to 0.01, and the RANSAC iteration number parameter is set to 1000.

[0030] Compared with the prior art, the beneficial effects of the present invention include: The present invention provides a method for obtaining the block size of blasted ore, including: the VCCS algorithm improved based on point cloud curvature and the CPC algorithm improved based on edge extraction. This method uses the VCCS algorithm improved based on point cloud curvature to segment data to obtain the first data, then uses the CPC algorithm improved based on edge extraction to cluster the first data to obtain the second data, and finally calculates the blasted block size according to the processed second data. This method improves the traditional supervoxel segmentation algorithm and achieves better segmentation results for the blasted point cloud data; at the same time, the VCCS algorithm improved based on point cloud curvature is used to reflect the concavity and convexity of the surface of the blasted ore, improving the boundary accuracy of the segmentation results; finally, the CPC algorithm improved based on edge extraction is used to reduce the problem of insensitivity to edge information, ultimately improving the accuracy of ore recognition, so as to accurately obtain the size of the blasted block size. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the local coordinate system m-XYM in Embodiment 1 of the present invention;

[0032] Figure 2 It is a schematic diagram of a triangle composed of an osculating circle, adjacent points, and a normal vector in Embodiment 1 of the present invention.

[0033] Figure 3 It is a display of the clustering result in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] This specific embodiment provides a method for obtaining the block size of blasted ore, including the following steps:

[0035] S1. Segment the cloud data using the VCCS algorithm improved based on point cloud curvature: Establish a voxelized grid, substitute the point cloud curvature into the supervoxel segmentation process, set the weight value parameter set before supervoxel segmentation as the curvature weight value, and then use the VCCS algorithm that performs voxelization processing on the point cloud data based on an octree and uses the K-means clustering algorithm for segmentation to segment the cloud data of the blasted points from a three-dimensional level to obtain the first segmented data;

[0036] S2. Strengthen the ore boundary information based on the edge extraction algorithm and then use the CPC algorithm to cluster the first data to obtain the second data; The input data of this algorithm is based on the first data segmented by the VCCS algorithm, and recursively segments through the concavity or convexity between two adjacent supervoxels to obtain the second data; In some embodiments, the maximum number of cuts of the CPC algorithm is 1000, the minimum size is set to 0.1, the minimum score that must be reached for cutting is set to 0.01, and the RANSAC iteration number parameter is set to 1000;

[0037] S3. Output the second data of each piece of ore and calculate the point cloud diameter. The size of the ore lump is obtained by calculating the three-dimensional spatial distance between the two farthest points in the second data of each clustering result.

[0038] In some embodiments, the cloud data is obtained by the following steps:

[0039] Use a three-dimensional laser scanner to perform multi-station scanning and acquisition of the cloud data before and after the mine blasting, and process the acquired cloud data with RiSCAN PRO or CloudCompare software to obtain it.

[0040] In some embodiments, in step S3, the three-dimensional spatial distance between the two farthest points is obtained by the following steps:

[0041] S31. Randomly select a point in the ore as the starting point A, select other adjacent points as the ending point B, calculate the distance between point A and point B, denoted as dis1, and record the starting point A, the ending point B, and the current distance Dis at this time. At this time, Dis = dis1;

[0042] S32. Replace any one of the two points A and B with an adjacent point C of these two points, and then calculate the distance between the un-replaced point A or point B and point C, denoted as dis2;

[0043] S33. Compare dis1 and dis2; if dis1 is greater than dis2, then continue to maintain the current point pair and distance, that is, the recorded point pair at this time is still the starting point A, the ending point B, and the distance dis1; if dis2 is greater than dis1, then the recorded point pair and distance should be updated in time at this time, and Dis = dis2 at this time; repeat the above steps until all points in the ore point cloud are traversed. As the traversal process continues, the calculated distance is gradually converging to the three-dimensional spatial distance between the two farthest points.

[0044] In some embodiments, in step S2, the strengthening of the ore boundary information based on the edge extraction algorithm includes: First, calculate the surface normal of each point in the muck pile points, make the tangent plane of this point according to the point to be calculated and its normal vector, then construct a local coordinate system with this point as the coordinate center, calculate the included angle between the vector from other points to this point and the established coordinate axis in the clockwise direction, compare the maximum value of the difference between adjacent included angles and the angle threshold. If the maximum value of the difference is greater than the angle threshold, then this point is an edge point. Repeat the above steps to calculate each point in the muck pile, and finally realize edge extraction.

[0045] In some embodiments, the point cloud curvature is obtained by the following steps:

[0046] Local fitting of normal curvature: Assume that there are n neighboring points for any point m in the muck pile point cloud, and the normal vector of the i-th neighboring point Pi is N i , taking point m as the origin, and using the orthogonal unit vectors X, Y of the coordinates of point m and its unit normal vector M as the coordinate axes, establish the local coordinate system m-XYM of m. In this coordinate system, the neighboring point P i has coordinates (x i , y i , z i ), and the normal vector N i of the neighboring point has coordinates (n x，i , n y，i , n z，i ). Then, the normal curvature passing through point m can be calculated by the osculating circle as

[0047] According to Euler's formula, the relationship between normal curvature and principal curvature is obtained:

[0048]

[0049] In the above formula, is the angle between the tangent of the normal section passing through point m and point p i and the principal direction; q1 and q2 are the two principal curvatures of point m.

[0050] In some embodiments, the formula for the normal curvature of point m relative to the neighboring point p i is as follows:

[0051]

[0052] In the formula, the α angle is the angle between the normal vector M of point m and mp i , the β angle is the angle between the normal vector M of point m and the normal vector N i of the neighboring point. The coordinates of the neighboring point P i are (x i , y i , z i ), and the coordinates of the normal vector N i of the neighboring point are (n x，i , n y，i , n z，i ), corresponding to x i , y i , n y,i , n x,i in the formula respectively.

[0053] Before obtaining the relationship between normal curvature and principal curvature according to Euler's formula, it also includes: least square fitting of Euler's equation:

[0054] Assume that the XY coordinates and the normal vector M at the given point m are respectively:

[0055]

[0056] Assume further that point m lies on plane L with normal vector M. Let e1 and e2 be the principal directions at point m, and the corresponding principal curvatures be k1 and k2. Let the parameter be the angle between vectors e1 and e2, and i be the angle between vector X and vector mP.

[0057] In some embodiments, in step S1, before adopting the VCCS algorithm, a voxelized grid is further established, and the parameters of the grid are set as follows: the voxel resolution Rvoxel = 0.025, the grid fixed resolution Rseed = 0.1, the distance weight Ws = 0.4, the normal vector weight Wn = 0.3. After the grid is set up, the VCCS algorithm is run to start the segmentation.

[0058] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0059] Embodiment 1

[0060] This embodiment provides a method for obtaining the block size of mucked ore, including the following steps:

[0061] S0. Collect cloud data through a laser scanner: In order to obtain three-dimensional cloud data, first use a three-dimensional laser scanner to perform multi-station scanning and collection of the cloud data before and after the mine blasting, and use RiSCAN PRO or CloudCompare software to process the collected cloud data. First, roughly delimit the range of the mucked ore point cloud data required for the experiment, and then use the cropping tool to crop and delete other redundant areas; then use the software to automatically identify and delete the noise point data within the delimited area after setting the parameters. For those noise point clouds that cannot be recognized by the software, a combination of manual visual inspection can be used for manual deletion; after performing the above operations on the data of multiple stations of the same mucked ore, finally, according to the steps of point cloud stitching (setting the reference station, adjusting the position and attitude of the point cloud, rough stitching, fine stitching, etc.), the multi-station point cloud data is stitched into a complete mucked ore point cloud data for use in the subsequent point cloud segmentation algorithm steps;

[0062] It should be noted that some ores with small block sizes are not easily able to clearly exhibit the characteristics of the ore during edge extraction, and the algorithm is unable to correctly identify them, affecting the subsequent calculation of the block size. The quality of the cloud data of the blasted ore heap should be improved by reducing the distance between the scanner and the blasted heap, enhancing the resolution of the scanner, increasing multi-station scanning, etc.;

[0063] S1. VCCS algorithm for segmenting cloud data based on point cloud curvature improvement: Establish a voxelized grid, substitute the point cloud curvature into the supervoxel segmentation process, and establish a voxelized grid. The parameter settings of the grid are as follows: voxel resolution Rvoxel = 0.025, grid fixed resolution Rseed = 0.1, distance weight Ws = 0.4, normal vector weight Wn = 0.3. After the grid is set, substitute the point cloud curvature into the supervoxel segmentation process, set the weight value parameter set before supervoxel segmentation as the curvature weight value, and then use the VCCS algorithm that performs voxelization processing on the point cloud data based on octree and uses the K-means clustering algorithm for segmentation to segment the cloud data of the blasted heap points from the three-dimensional level to obtain the first segmented data. Run the VCCS algorithm to start segmentation, and use the VCCS algorithm to segment the cloud data of the blasted heap points from the three-dimensional level to obtain the first segmented data; This step of processing improves the accuracy of supervoxel segmentation to better determine the edges of overlapping ores; The improved VCCS algorithm that uses point cloud curvature instead of cloud color for supervoxel segmentation can reflect the concavity and convexity of the ore surface and has higher segmentation accuracy;

[0064] The boundary information of the ore is related to the degree of curvature of its surface, that is, curvature. To calculate the curvature of a point, first construct a normal cross-section circle, and then calculate the positional relationship between the target point and adjacent points and the normal vector estimation to obtain the point cloud curvature (i.e., the principal curvature). The specific process is as follows:

[0065] S11. Local fitting of normal curvature: Assume that any point m in the blasted heap point cloud has n adjacent points, and the normal vector of the i-th adjacent point Pi is N i , taking point m as the origin, using the orthogonal unit vectors X, Y of the coordinates of point m and its unit normal vector M as the coordinate axes, establish the local coordinate system m-XYM of m. In this coordinate system, the coordinates of the adjacent point P i are (x i , y i , z i ), and the coordinates of the normal vector N i of the adjacent point are (n x，i , n y，i , n z，i ). Then, the normal curvature of point m can be calculated through its osculating circle as The geometric relationships between the variables are as follows Figure 1 and 2As shown in; among which, Figure 1 is the local coordinate system m-XYM. Figure 2 is a triangle composed of the osculating circle, the neighboring point, and the normal vector. The normal curvature formula of point m with respect to the neighboring point p i is as follows:

[0066]

[0067] In the above formula, the α angle is the angle between the normal vector M of point m and mp i and the β angle is the angle between the normal vector M of point m and the normal vector N of the neighboring point i ; the coordinates of the neighboring point P i are (x i , y i , z i ), and the coordinates of the normal vector N of the neighboring point i are (n x，i , n y，i , n z，i ), corresponding to x i , y i , n y,i , n x,i ;

[0068] S12, Least Squares Fitting of Euler's Equation: When calculating the normal curvature, since it is solved in the newly established local coordinate system, it involves the coordinate transformation problem between the global coordinates of the neighboring point and the local coordinates. Assume that the XY coordinates and the normal vector M at the given point m are respectively:

[0069]

[0070] Assume again that point m is on the plane L, its normal vector is M, let e1 and e2 be the principal directions at point m, and the corresponding principal curvatures are k1 and k2. Let the parameter be the angle between the vectors e1 and e2, be the angle between the vector X and the vector mP i (the projection of mP i on the plane L), where can be calculated using the local coordinates of p i . According to Euler's formula, the relationship between the normal curvature and the principal curvature can be obtained:

[0071]

[0072] In the above formula, is the normal curvature of point m passing through point p iThe angle between the tangent of the normal section and the principal direction; q1 and q2 are the two principal curvatures of point m. The principal curvature value of the point cloud can be calculated according to the above formula. The algorithm based on the point cloud curvature has a better effect on the segmentation of the muckpile ore.

[0073] S2. After strengthening the ore boundary information based on the edge extraction algorithm, the CPC algorithm is used to cluster the first data to obtain the second data; the input data of this algorithm is based on the first data segmented by the VCCS algorithm, and the second data is obtained by recursively segmenting through the concavity or convexity between two adjacent supervoxels; the maximum number of cuts of the CPC algorithm is 1000, the minimum size is set to 0.1, the minimum score that must be achieved for cutting is set to 0.01, and the RANSAC iteration number parameter is set to 1000; after the CPC algorithm completes clustering, it reflects the spatial positions between the ores. The clustering results are shown in Figure 3 :

[0074] According to Figure 3 it can be seen that the clustering effect is clear and there is no situation where multiple small pieces of ore are contained in a large piece of ore in the clustering result due to the different concavities and convexities of the ore surface. On the contrary, we can Figure 3 find the boundaries of each ore in the muckpile well, with a better ore recognition effect and higher recognition accuracy;

[0075] Due to the overly complex shape, size, and distribution of the ores in the muckpile, using the traditional CPC algorithm for clustering is prone to incorrect clustering groups. Therefore, before the CPC algorithm, the ore boundary information is strengthened based on the edge extraction algorithm, and then the CPC algorithm is used to cluster the cloud data. The edge extraction and strengthening process is as follows: First, calculate the surface normal of each point in the muckpile points. Make the tangent plane of this point according to the point to be calculated and its normal vector. Then, construct a local coordinate system with this point as the coordinate center, calculate the angle between the vector from other points to this point and the established coordinate axis in the clockwise direction, and judge the size of the maximum value of the difference between adjacent angles and the set angle threshold. In this embodiment, the angle threshold is π / 2. If the maximum value of the difference is greater than the angle threshold, then this point is an edge point. Repeat the above steps to calculate each point in the muckpile, and finally achieve edge extraction;

[0076] In this embodiment, the CPC algorithm needs to be optimized first. The number of clouds should not be too small, otherwise an ore cannot be formed, and data with too few clouds should be deleted as discrete noise points. Then, the point cloud is compared with the muck pile range, and the point cloud beyond the range is deleted. After completing the above steps, the calculated block size is compared with the maximum ore size. If the block size is greater than the maximum size, the edge extraction algorithm mentioned in step S3 should be iterated again to strengthen the boundary information of each ore in the ore group, re-cluster using the CPC algorithm, and repeat the above steps until the calculated ore block size is less than the maximum ore size. If the block size is less than the maximum size, the result can be output.

[0077] S3. Calculate the block size of the muck pile ore: Output the second data of each ore in blocks and calculate the point cloud diameter. The size of the ore block is obtained by calculating the three-dimensional space distance (i.e., Euclidean distance, also called Euclidean distance) between the two farthest points in each clustering result. Assume that there are any two points A(a1, a2, a3) and B(b1, b2, b3) in the point cloud data, then the distance between point A and point B is:

[0078]

[0079] The process of the algorithm to find the two farthest points in the ore point cloud mainly includes three steps:

[0080] S31. First, randomly select a point in the ore as the starting point A, select other adjacent points as the ending point B, calculate the distance between point A and point B, denoted as dis1, and record the starting point A, the ending point B, and the current distance Dis at this time. At this time, Dis = dis1;

[0081] S32. Replace any one of the two points with their adjacent point C, and then calculate the distance between the un-replaced point A or point B and point C (if the starting point A is replaced, then calculate the distance between point B and point C; if the ending point B is replaced, then calculate the distance between point A and point C), denoted as dis2;

[0082] S33. Finally, compare dis1 and dis2; if dis1 is greater than dis2, then continue to maintain the current point pair and distance, that is, the recorded point pair is still the starting point A, the ending point B, and the distance dis1 at this time; if dis2 is greater than dis1, then the recorded point pair and distance should be updated in time. At this time, Dis = dis2. Repeat the above steps until all points in the ore point cloud are traversed. As the traversal process continues, the calculated distance is gradually converging to the true block size of the ore. That is to say, the coordinates of the points on the ore are known, and by calculating the spatial distance between the farthest points in each group, the size of the ore block can be obtained.

[0083] Other beneficial effects of the present invention include:

[0084] 1) By using a three-dimensional laser scanner to collect cloud data, and by methods such as improving the scanner resolution and shortening the distance between the scanner and the muck pile, more comprehensive and detailed cloud data of the muck pile can be obtained, while ensuring that even smaller ores can be identified;

[0085] 2) Using the VCCS algorithm improved based on point cloud curvature to segment cloud data. Calculating the curvature before the traditional VCCS algorithm can more intuitively reflect the boundary information of the ore and the concavity and convexity of the surface, etc., making the accuracy of the final cloud data segmentation result higher;

[0086] 3) Using the CPC algorithm improved based on edge extraction to cluster the data. Performing edge extraction as a preprocessing before the traditional CPC algorithm clusters the cloud data is beneficial for the algorithm to accurately find the boundary information of the ore, beneficial for the identification of the muck pile, and at the same time solves the problem of incorrect clustering existing in the traditional algorithm;

[0087] 4) When calculating the block size of the ore muck pile, the CPC algorithm based on edge extraction is optimized again. After optimization, only corresponding parameters need to be set before the algorithm starts to run, and after inputting the processed cloud data, the algorithm can run by itself without any manual intervention in the middle, achieving the automation of ore block size calculation and greatly improving the convenience.

[0088] The specific implementation manners of the present invention described above do not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and deformations made according to the technical concept of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A method for obtaining the block size of blasted ore, characterized in that, It includes the following steps: S1. Segment the cloud data by the VCCS algorithm improved based on point cloud curvature: Establish a voxelized grid, substitute the point cloud curvature into the supervoxel segmentation process, set the weight value parameter set before supervoxel segmentation as the curvature weight value, and then use the VCCS algorithm that performs voxelization processing on the point cloud data based on an octree and uses the K-means clustering algorithm for segmentation to segment the cloud data of the muck pile points from a three-dimensional level to obtain the first segmented data; S2. Strengthen the ore boundary information based on the edge extraction algorithm and then use the CPC algorithm to perform clustering processing on the first data to obtain the second data; The input data of this algorithm is based on the first data segmented by the VCCS algorithm, and the second data is obtained by recursively segmenting through the concavity or convexity between two adjacent supervoxels; S3. Calculate the muck pile ore size: Output the second data of each piece of ore and calculate the point cloud diameter. The size of the ore size is obtained by calculating the three-dimensional space distance between the two farthest points in the second data of each clustering result; In step S1, the point cloud curvature is obtained by the following steps: Local fitting of normal curvature: Assume that there are n neighboring points for any point m in the muck pile point cloud, and the normal vector of the i-th neighboring point Pi is N i , taking point m as the origin, and using the orthogonal unit vectors X, Y of the coordinates of point m and its unit normal vector M as the coordinate axes, establish the local coordinate system m-XYM of m. In this coordinate system, the neighboring point P i has coordinates (x i , y i , z i ), and the normal vector N i of the neighboring point has coordinates (n x，i , n y，i , n z，i ). Then, the normal curvature passing through point m can be calculated as Obtain the relationship between the normal curvature and the principal curvature according to Euler's formula: In the above formula, is the angle between the tangent of the normal section passing through point m and point p i and the principal direction; q1 and q2 are the two principal curvatures of point m; The normal curvature formula of point m with respect to the adjacent point p i is as follows: In the formula, the α angle is the angle between the normal vector M of point m and m p i and the β angle is the angle between the normal vector M of point m and the normal vector N of the adjacent point i The coordinates of the adjacent point P i are (x i , y i , z i ), and the coordinates of the normal vector N of the adjacent point i are (n x，i , n y，i , n z，i ), corresponding to x i , y i , n y,i , n x,i in the formula respectively.

2. The method for obtaining the block size of mucked ore according to claim 1, characterized in that In step S1, the cloud data is obtained by the following steps: Use a three-dimensional laser scanner to perform multi-station scanning and acquisition of the cloud data before and after the mine blasting, and process the acquired data with RiSCAN PRO or CloudCompare software to obtain it.

3. The method for obtaining the fragment size of mucked ore according to claim 1, characterized in that, In step S3, the three-dimensional space distance between the two farthest points is obtained by the following steps: S31. Randomly select a point in the ore as the starting point A, select other adjacent points as the ending point B, calculate the distance between point A and point B, denoted as dis1, and record the starting point A, the ending point B, and the current distance Dis at this time. At this time, Dis = dis1; S32. Replace any one of the two points A and B with an adjacent point C of these two points, and then calculate the distance between the un-replaced point A or point B and point C, denoted as dis2; S33. Compare dis1 and dis2; if dis1 is greater than dis2, then continue to maintain the current point pair and distance, that is, the recorded point pair at this time is still the starting point A, the ending point B, and the distance dis1; if dis2 is greater than dis1, then the recorded point pair and distance at this time should be updated in time. At this time, Dis = dis2; Repeat the above steps until all points in the ore point cloud are traversed. As the traversal process continues, the calculated distance is gradually converging to the three-dimensional space distance between the two farthest points.

4. The method for obtaining the fragment size of mucked ore according to claim 1, characterized in that, In step S2, the enhancement of the ore boundary information based on the edge extraction algorithm includes: First, calculate the surface normal of each point in the muck pile points, make the tangent plane of this point according to the point to be calculated and its normal vector, then build a local coordinate system with this point as the coordinate center, calculate the angle between the vector from other points to this point and the established coordinate axis in the clockwise direction, compare the maximum value of the difference between adjacent angles and the angle threshold. If the maximum value of the difference is greater than the angle threshold, then this point is an edge point. Repeat the above steps to calculate each point in the muck pile, and finally realize edge extraction.

5. The method for obtaining the block size of blasted ore according to claim 1, wherein Before obtaining the relationship between the normal curvature and the principal curvature according to Euler's formula, it also includes: the least squares fitting of Euler's equation: Assume that the XY coordinates and the normal vector M at the given point m are respectively: Assume again that point m lies on plane L with normal vector M. Let e1 and e2 be the principal directions at point m, and the corresponding principal curvatures be k1 and k2. Let the parameter be the angle between vectors e1 and e2, and i be the angle between vector X and vector mP.

6. The method for obtaining the block size of blasted muck according to claim 1, characterized in that In step S1, before adopting the VCCS algorithm, it also includes establishing a voxelized grid, and the parameters of the grid are set as: the voxel resolution Rvoxel = 0.025, the grid fixed resolution Rseed = 0.1, the distance weight Ws = 0.4, the normal vector weight Wn = 0.

3. After the grid is set up, run the VCCS algorithm to start segmentation.

7. The method for obtaining the block size of mucked ore according to claim 1, wherein In step S2, the maximum number of cuts of the CPC algorithm is 1000, the minimum size is set to 0.1, the minimum score that the cut must reach is set to 0.01, and the RANSAC iteration number parameter is set to 1000.

Images