A method for calculating the burst size of multi-scale-dimensional features of 3D laser point clouds

Abstract

This invention discloses a method for calculating the block size of blasted piles based on multi-scale-dimensional features of three-dimensional laser point clouds, applied in the field of basic geographic information technology. The method includes the following steps: S1. Extraction of multi-scale-dimensional features of the three-dimensional laser point cloud; S2. An improved Euclidean segmentation method for the multi-scale-dimensional features; S3. Obtaining the calculation results of the blasted pile block size; S4. Obtaining the block size distribution of the blasted pile. This invention realizes the application of three-dimensional laser scanning technology to the analysis of blasted pile block size after mine blasting, and optimizes ore point cloud identification by combining multi-scale-dimensional features of the point cloud with normal vector features, thereby improving the calculation accuracy of blasted pile block size.

Description

Technical Field

[0001] This invention relates to the field of basic geographic information technology, and more specifically to a method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds. Background Technology

[0002] Against the backdrop of advocating the development of green and smart mines, energy utilization efficiency and post-blast environmental impact are crucial issues in open-pit mine bench blasting. Maximizing the effectiveness of open-pit mine bench blasting positively impacts both economic returns and production costs. Burst size is a key indicator of blasting effectiveness, reflecting not only the rationality of various parameters but also significantly influencing post-blast ore transport efficiency and economic benefits. With the development of 3D laser scanning technology, the accuracy and efficiency of mine surveying have greatly improved. Compared to traditional methods, 3D laser scanning offers advantages such as high precision, high efficiency, and intelligence, enabling more detailed and comprehensive reproduction of mine scenes and blasting conditions, thus gaining widespread application in mine surveying. Calculating burst size using 3D laser point cloud data improves accuracy, thereby helping to control blasting effects and mining costs. Therefore, research on burst size calculation based on 3D laser point cloud data is of significant importance.

[0003] Therefore, proposing a method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds to solve the difficulties existing in the prior art is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds, which is used to solve the technical problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds includes the following steps:

[0007] S1. Extraction of multi-scale-dimensional features of 3D laser point cloud: Based on the preprocessed 3D laser point cloud data of the burst pile, multi-scale-dimensional features of the point cloud are extracted by a multi-scale-dimensional feature extraction algorithm to obtain the results of the extracted multi-scale-dimensional features.

[0008] S2. Improved Euclidean segmentation method based on multi-scale-dimensional features: Based on the extracted multi-scale-dimensional features, a multi-feature fusion method is used to combine the multi-scale-dimensional features with the normal vector features as the segmentation criterion for Euclidean segmentation.

[0009] S3. Calculation results of the block size of the blasted ore: Based on the locally improved parameter adaptive method, the block size of the blasted ore is calculated by the maximum distance algorithm of the point cloud, the ore identification is judged, and the local ore identification error is optimized. Finally, the point cloud of the blasted ore with high identification accuracy and the block size calculation results of the blasted ore are output.

[0010] S4. Block size distribution of the blast pile: Based on the block size calculation results of the blast pile ore, the ore is classified and statistically analyzed according to the ore discharge and feed diameters of different specifications of crushing machinery in the mining beneficiation process, and the block size distribution of the entire blast pile is obtained.

[0011] Optionally, the results of the multi-scale-dimensional features extracted in S1 can be used to improve subsequent point cloud segmentation algorithms.

[0012] Optionally, the multi-scale-dimensional features of the point cloud in S1 are a multi-scale measure of the point cloud dimension defined by the neighborhood of each point.

[0013] Optionally, the multi-scale-dimensional feature extraction algorithm in S1 selects training samples for the point cloud data to be segmented, and finds the optimal scale combination of the classifier based on the training samples. The final classification result of the classifier is affected by the comprehensive influence of local dimensional information at multiple scales, so that the target classification has the maximum separability.

[0014] Optionally, the overall implementation flow of the multi-scale-dimensional feature extraction algorithm in S1 is as follows:

[0015] First, PCA is used to analyze the dimensional features of point clouds of training samples at different scales. Second, a classifier is constructed using support vector machines. The optimal combination of classification scales is found based on the multi-scale-dimensional features of the training samples to generate the optimal classification hyperplane. Finally, the dimensional feature information of the bursty point cloud data is extracted.

[0016] Optionally, in S2, based on the results of the extracted multi-scale-dimensional features, the ore in the blast pile area is divided into fine crushed ore and large ore according to different mining area loading equipment and coarse crusher specifications.

[0017] Optionally, in S2, normal vector feature extraction is a normal estimation of the local fitted plane of a point in the point cloud. The normal vector feature can reflect the direction of the local plane of the point cloud.

[0018] Optionally, the process of the locally improved parameter adaptive method in S3 is as follows: First, calculate the ore size of the ore segmentation result before performing local improvement; then, manually measure the size of the largest ore in the blast pile area, and by comparing the size of each ore piece with the size of the largest ore in the blast pile area, identify the ore with adhesion, and then perform local optimization of the ore identification result; finally, for the ore point cloud that needs to optimize the identification result, the improved Euclidean segmentation method with multi-scale-dimensional features needs to be used again to redetermine the normal vector angle threshold and perform local secondary segmentation.

[0019] Optionally, the distribution of blast pile block size in S4 is usually achieved by classifying and statistically analyzing the ore in the blast pile according to different specifications, calculating the mass of each block size, and then calculating the percentage of each block size in the total mass, thereby obtaining the ore classification specifications.

[0020] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method for calculating the burst size of multi-scale-dimensional features of three-dimensional laser point clouds, and its beneficial effects are as follows:

[0021] 1) This invention realizes the application of three-dimensional laser scanning technology in the analysis of blasted pile size after mine blasting. It combines multi-scale-dimensional features of point cloud and normal vector features to optimize ore point cloud identification, improves the calculation accuracy of blasted pile size, and overcomes the shortcomings of traditional blasted pile size calculation methods such as low accuracy and large workload.

[0022] 2) Compared to traditional methods for calculating the size of blasted ore deposits, the method based on 3D laser scanning technology offers advantages such as high security, fast data acquisition, and high efficiency in calculating ore size. Furthermore, it addresses the issue of inaccurate blasted ore size calculations due to the low accuracy of traditional 3D laser point cloud segmentation algorithms in identifying blasted ore deposits.

[0023] 3) By adopting a multi-feature fusion approach, the multi-scale-dimensional features of point clouds are combined with normal vector features as the segmentation criteria for point cloud segmentation algorithms, which improves the traditional segmentation algorithm and increases the recognition accuracy of blasted ore and the calculation accuracy of blasted block size.

[0024] 4) From the acquisition of 3D laser point cloud data of blast piles to the display of blast pile ore identification results and block size calculation results, the advantages of existing technologies are drawn upon while overcoming their shortcomings. Practical application has proven that it significantly improves the work efficiency of the blast pile block size analysis process and the data accuracy of the calculation results. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0026] Figure 1 A flowchart of the method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds provided by the present invention;

[0027] Figure 2 A schematic diagram of the method flow is provided for embodiments of the present invention;

[0028] Figure 3 This is a schematic diagram of the multi-scale-dimensional feature extraction process for bursty point clouds provided in an embodiment of the present invention;

[0029] Figure 4 The effect diagram of the preprocessed point cloud provided in the embodiment of the present invention;

[0030] Figure 5 The image shows the effect of multi-scale-dimensional feature extraction on the burst point cloud provided in the embodiment of the present invention.

[0031] Figure 6 The image shows the effect of ore identification after using the point cloud of the explosive pile provided in the embodiment of the present invention.

[0032] Figure 7 This is a distribution map of ore size gradation at the point cloud of the blast pile provided in an embodiment of the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] See Figure 1 As shown, this invention discloses a method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds, including the following steps:

[0035] S1. Extraction of multi-scale-dimensional features of 3D laser point cloud: Based on the preprocessed 3D laser point cloud data of the burst pile, multi-scale-dimensional features of the point cloud are extracted by a multi-scale-dimensional feature extraction algorithm to obtain the results of the extracted multi-scale-dimensional features.

[0036] S2. Improved Euclidean segmentation method based on multi-scale-dimensional features: Based on the extracted multi-scale-dimensional features, a multi-feature fusion method is used to combine the multi-scale-dimensional features with the normal vector features as the segmentation criterion for Euclidean segmentation.

[0037] S3. Calculation results of the block size of the blasted ore: Based on the locally improved parameter adaptive method, the block size of the blasted ore is calculated by the maximum distance algorithm of the point cloud, the ore identification is judged, and the local ore identification error is optimized. Finally, the point cloud of the blasted ore with high identification accuracy and the block size calculation results of the blasted ore are output.

[0038] S4. Block size distribution of the blast pile: Based on the block size calculation results of the blast pile ore, the ore is classified and statistically analyzed according to the ore discharge and feed diameters of different specifications of crushing machinery in the mining beneficiation process, and the block size distribution of the entire blast pile is obtained.

[0039] Specifically, crushing machinery includes coarse crushers, fine crushers, and ball mills.

[0040] Furthermore, the results of the multi-scale-dimensional features extracted in S1 are used to improve subsequent point cloud segmentation algorithms.

[0041] Furthermore, the multi-scale-dimensional features of the point cloud in S1 are a multi-scale measure of the point cloud dimension defined by the neighborhood of each point.

[0042] Specifically, leveraging the principle that point clouds exhibit different dimensional characteristics at different scales, principal component analysis (PCA) is performed on each neighboring point in the neighborhood sphere within a Cartesian coordinate system. The eigenvalues ​​output by the PCA are compared to determine whether the point cloud can be interpreted as a one-dimensional line state, a two-dimensional surface state, or a three-dimensional volume state.

[0043] Furthermore, the multi-scale-dimensional feature extraction algorithm in S1 selects training samples for the point cloud data to be segmented, and finds the optimal scale combination of the classifier based on the training samples. The final classification result of the classifier is affected by the comprehensive influence of local dimensional information of multiple scales, so that the target classification has the maximum separability.

[0044] Furthermore, the overall implementation process of the multi-scale-dimensional feature extraction algorithm in S1 is as follows:

[0045] First, PCA is used to analyze the dimensional features of point clouds of training samples at different scales. Second, a classifier is constructed using support vector machines. The optimal combination of classification scales is found based on the multi-scale-dimensional features of the training samples to generate the optimal classification hyperplane. Finally, the dimensional feature information of the bursty point cloud data is extracted.

[0046] Furthermore, in S2, based on the results of the extracted multi-scale-dimensional features, the ore in the blast pile area is divided into fine crushed ore and large ore according to different mining area loading equipment and coarse crusher specifications.

[0047] Specifically, based on the capacity volume V of the loading equipment bucket, the minimum allowable feed size D of the primary crusher can be determined using the formula: Obtain the defining dimension d between fine crushed ore and large ore. Assume the loading equipment is a WK-55 mining excavator with a bucket capacity of 36-76 cubic meters, the coarse crusher is a PE250×1200 jaw crusher, the size of large ore is greater than 1.2 meters, and the size of fine crushed ore is less than or equal to 1.2 meters.

[0048] Specifically, fine fragments of ore are represented by three-dimensional feature points, while large chunks of ore are represented by two-dimensional feature points. In the point clouds of these two types of ore, the difference in normal vectors between fine fragments and large chunks is relatively small. Therefore, under the original Euclidean clustering traversal search condition (searching for neighboring points based on the Euclidean distance threshold), for the point to be classified, an additional classification criterion is added: if the point's multi-scale-dimensional features are two-dimensional, the point's normal vector differences are relatively small, and the original Euclidean segmentation condition is applied for classification; if the point's multi-scale-dimensional features are three-dimensional, a normal vector condition is added to the original Euclidean segmentation condition for classification.

[0049] Furthermore, in S2, normal vector feature extraction is a normal estimation of the local fitted plane of a point in the point cloud. The normal vector feature can reflect the direction of the local plane of the point cloud.

[0050] Furthermore, the process of the locally improved parameter adaptive method in S3 is as follows: First, calculate the ore size of the ore segmentation result before performing local improvement; then, manually measure the size of the largest ore in the blast pile area, and by comparing the size of each ore piece with the size of the largest ore in the blast pile area, identify the ore with adhesion, and then perform local optimization of the ore identification result; finally, for the ore point cloud that needs to optimize the identification result, the improved Euclidean segmentation method with multi-scale-dimensional features needs to be used again to redetermine the normal vector angle threshold and perform local secondary segmentation.

[0051] Specifically, the parameter adaptive adjustment algorithm mainly aims to solve the following problems: In different blast piles, the different stacking conditions of the ore will lead to the diversity of the angle between the normal vectors at the contact positions of the ore. Relying solely on the single discrimination and segmentation of the normal vector angle cannot be well adapted to the point cloud data of different blast piles. As a result, the segmentation algorithm applied to certain blast pile areas will result in the phenomenon of local ore point cloud adhesion.

[0052] Furthermore, the distribution of blast pile block size in S4 is usually determined by classifying and statistically analyzing the ore in the blast pile according to different specifications, calculating the mass of each block size, and then calculating the percentage of each block size in the total mass, thereby obtaining the ore classification specifications.

[0053] Specifically, the grading specifications of ores vary depending on the type of mineral, the industrial use of the ore, and the method of calculating the block size.

[0054] In a specific embodiment, this embodiment implements a method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds. See the flowchart for details. Figure 2 As shown, firstly, a RIEGL_VZ1000 3D laser scanner was selected to collect burst pile data. The collection area included the copper plant and Fujiawu mining areas of Dexing Copper Mine. Then, for the collected burst pile data, RiSCANPro software was used to perform preprocessing operations on the point cloud data, such as multi-station registration, cropping, and denoising. Finally, the preprocessed burst pile point cloud data was obtained, as shown in the figure. Figure 4 As shown.

[0055] like Figure 3 As shown, the steps for extracting multi-scale-dimensional features from the preprocessed point cloud data are as follows: First, the point cloud is manually cropped to obtain large and small ore point cloud samples. Then, PCA is used to analyze the point cloud dimensional features of the training samples at different scales. Based on this, a support vector machine is used to construct a classifier. The optimal combination of classification scales is found based on the multi-scale-dimensional features of the training samples, generating the optimal classification hyperplane. Finally, the point cloud is classified based on the optimal classification hyperplane, thus extracting the dimensional feature information of the point cloud data. The extraction results are shown below. Figure 5 As shown.

[0056] Based on the results of the multi-scale-dimensional feature extraction algorithm, the ore in the blast pile area was coarsely classified into fine-grained ore (3D feature points) and large-scale ore (2D feature points). In the point clouds of the two types of ore, the difference in normal vectors is relatively large for fine-grained ore and relatively small for large-scale ore. Therefore, under the original Euclidean clustering traversal search condition (searching for neighboring points based on the Euclidean distance threshold), for the point to be classified, an additional classification criterion is added: if the multi-scale-dimensional feature of the point is a 2D feature, the point is generally in a situation where the difference in normal vectors is relatively small, and the original Euclidean segmentation condition is applied for classification; if the multi-scale-dimensional feature of the point is a 3D feature, a normal vector condition is added to the original Euclidean segmentation condition for classification.

[0057] To address local segmentation errors in some ore heap areas, an adaptive parameter adjustment algorithm is needed. First, before local improvements, the ore size in the segmented ore results is calculated. Then, the size of the largest ore in the heap area is manually measured. By comparing the size of each ore piece with the size of the largest ore in the heap area, ore exhibiting adhesion can be identified, allowing for local optimization of the ore identification results. Finally, for the ore point cloud requiring further optimization, the normal vector angle threshold in the algorithm is redefined, and secondary local segmentation is performed. The final identification result of the heaped ore is as follows: Figure 6 As shown.

[0058] Based on the identification results and size calculation results of the ore in the burst pile, and using the crushing instruments in the mineral processing stage—PE250×1200 jaw crusher, MP800 cone crusher, and Φ7.32×10.68m dual-motor driven overflow ball mill—the ore in the burst pile was classified and statistically analyzed according to the following sizes: less than 0.05m, 0.05~0.1m, 0.1~0.2m, 0.2~0.3m, 0.3~0.4m, 0.4~0.6m, 0.6~0.8m, 0.8~1.0m, 1.0~1.2m, and greater than 1.2m. This yielded a size distribution map of the ore in the burst pile, as shown below. Figure 7 As shown;

[0059] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0060] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those 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 invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds, characterized in that, Includes the following steps: S1. Extraction of multi-scale-dimensional features of 3D laser point cloud: Based on the preprocessed 3D laser point cloud data of the burst pile, multi-scale-dimensional features of the point cloud are extracted by a multi-scale-dimensional feature extraction algorithm to obtain the results of the extracted multi-scale-dimensional features. S2. Improved Euclidean segmentation method based on multi-scale-dimensional features: Based on the extracted multi-scale-dimensional features, a multi-feature fusion method is used to combine the multi-scale-dimensional features with the normal vector features as the segmentation criterion for Euclidean segmentation. S3. Calculation results of the block size of the blasted ore: Based on the locally improved parameter adaptive method, the block size of the blasted ore is calculated by the maximum distance algorithm of the point cloud, the ore identification is judged, and the local ore identification error is optimized. Finally, the point cloud of the blasted ore with high identification accuracy and the block size calculation results of the blasted ore are output. S4. Block size distribution of the blast pile: Based on the block size calculation results of the blast pile ore, the ore is classified and statistically analyzed according to the ore discharge and feed diameters of different specifications of crushing machinery in the mining beneficiation process, and the block size distribution of the entire blast pile is obtained. The overall implementation process of the multi-scale-dimensional feature extraction algorithm in S1 is as follows: First, PCA is used to analyze the point cloud dimensional features of training samples at different scales. Second, a classifier is constructed using support vector machine. The optimal classification scale combination is found based on the multi-scale-dimensional features of the training samples to generate the optimal classification hyperplane. Finally, the dimensional feature information of the bursty point cloud data is extracted. In S2, normal vector feature extraction is a normal estimation of the local fitted plane of a point in the point cloud. The normal vector feature can reflect the direction of the local plane of the point cloud. The process of the locally improved parameter adaptive method in S3 is as follows: First, calculate the ore size of the ore segmentation result before performing local improvement; then, manually measure the size of the largest ore in the blast pile area, and by comparing the size of each ore piece with the size of the largest ore in the blast pile area, identify the ore with adhesion, and then perform local optimization of the ore identification result; finally, for the ore point cloud that needs to optimize the identification result, the improved Euclidean segmentation method with multi-scale-dimensional features needs to be used again to redetermine the normal vector angle threshold and perform local secondary segmentation.

2. The method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds according to claim 1, characterized in that, The results of the multi-scale-dimensional features extracted in S1 are used to improve the subsequent point cloud segmentation algorithm.

3. The method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds according to claim 1, characterized in that, The multi-scale-dimensional feature of the point cloud in S1 is a multi-scale measure of the point cloud dimension defined by the neighborhood of each point.

4. The method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds according to claim 1, characterized in that, The multi-scale-dimensional feature extraction algorithm in S1 selects training samples from the point cloud data that needs to be segmented, and finds the optimal scale combination for the classifier based on the training samples.

5. The method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds according to claim 1, characterized in that, Based on the extracted multi-scale-dimensional features in S2, the ore in the blasting area is divided into fine crushed ore and large ore according to different mining area loading equipment and coarse crusher specifications.

6. The method for calculating the burst density of multi-scale-dimensional features of three-dimensional laser point clouds according to claim 1, characterized in that, In S4, the distribution of blast pile block size is usually determined by classifying and statistically analyzing the ore in the blast pile according to different specifications, calculating the mass of each block size, and then calculating the percentage of each block size to the total mass, thereby obtaining the ore classification specifications.

Images