Fragmentation Calculation Method For Blast Muck Piles In Open-Pit Copper Mines Based On Concave-Convex Features Of Laser Point Cloud Data

Nowadays,many open-pit mines are based on two-dimensional image system to identify the ore of blast muck piles,but there are many disadvantages of the method.In order to obtain a higher accuracy result of fragmentation calculation,the paper uses3D laser scanning technology to scan the blast muck piles to obtain high-quality 3D point cloud data.The paper describes the features of the ore of blast muck piles from the 3D space domain.It also uses the point cloud segmentation algorithm and improves the shortcomings of algorithm to achieve accurate segmentation results of point cloud data of blast muck piles.In this paper,a fragmentation calculation method for blast muck piles in open-pit copper mine based on the concave-convex features of laser point cloud is proposed.The main work includes as the following:（1）The voxel cloud connectivity segmentation（VCCS）method based on the curvature of the point cloud is used to perform super voxel segmentation on the point cloud data of blast muck piles.The color difference between adjacent ores in the same blast muck pile is not obvious.When the original VCCS algorithm is used to segment the point cloud data of blast muck pile,the little influence is existed on the segmentation results by setting different color weights.In order to make the VCCS algorithm more suitable for the field of blast muck piles,the curvature of point cloud data,which can reflect the concave-convex degree of the ore surfaces of blast muck piles,is used to replace the color information for super voxel segmentation in the paper.Experiment results show that the curvature of point cloud is beneficial to improve the accuracy of ore identification.At the same time,the rational setting of curvature weight is discussed.After comparing many experimental results,the optimal curvature weight value of supervoxel segmentation results of blast muck piles is 0.7.（2）The constrained planar cuts（CPC）method based on edge extraction is used to cluster the super voxel segmentation results of blast muck pile.The CPC algorithm is performed by judging the concave-convex relationship between super voxels and uses the local constrained,directed random sample consensus（RANSAC）algorithm to perform hierarchical,recursive,and independent segmentation.Although the results of ore identification by CPC algorithm is good,due to the different shape,size and distribution of ores in blast muck pile,some wrong results are still obtained when using CPC algorithm to cluster the ore group whose ore boundary information is not obvious in blast muck pile.Therefore,in order to solve the problem,the paper proposes an improved algorithm based on edge extraction,and the experiment results prove that the ore groups which are wrongly clustered together can be separated accurately.At the same time,there is a certain influence on the fragmentation calculation of the final ore results by using the edge extraction algorithm.The paper also discusses the error size of the edge extraction iteration times for the fragmentation calculation results,and shows that it is most appropriate when the iteration times are set to 2-3 times.（3）The fragmentation of ores is calculated and the quality of blast muck piles is evaluated.After obtaining the high-precision identification results of ores of blast muck piles,it is necessary to quantitatively evaluate the accuracy of the clustering results.Three evaluation indexes of ore identification,accuracy,recall and F₁ score,are proposed in the paper.The effectiveness of the improved algorithm in the paper is quantitatively verified,and comparison with other common methods of blasting point cloud segmentation is also discussed.The experiment results prove that the accuracy of ore identification results obtained by the proposed algorithm in the paper is the highest.At the same time,in the aspect of ore fragmentation calculation,the paper optimizes the proposed algorithm,reduces manual intervention as much as possible,and accomplishes the automatic fragmentation calculation process of blast muck piles.The fragmentation measurement of ores is obtained by calculating the distance between two farthest points in the point cloud data contained in each clustering result in three-dimensional space,that is,Euclidean distance.At last,the paper put forward several evaluation indexes of the quality of blast muck piles,such as the number of ores,P₂₀,P₅₀,P₈₀,the size of the largest/smallest ores,the proportion of fine ores,etc.The paper also calculates the rate of bulk ores,and evaluates quantitatively the fragmentation of blast muck piles.By comprehensive analysis of various factors,it is concluded that the quality of the blast muck piles in the paper is good.