Point cloud segmentation method for rock pile particle size analysis based on edge detection and region growing

Rock fragmentation caused by blasting affects the productivity and efficiency of downstream operations (including processing and transportation). Analyzing the size composition of rock particles is essential for optimizing blasting design. Current methods for analyzing rock particle size rely on time-consuming and labor-intensive sieving experiments. This study proposes a method for automatically analyzing the size of rock pile particles. First, a point cloud of the rock pile is collected using a handheld laser scanner. Next, a rock particle point cloud edge detection algorithm based on adjacent features is proposed to obtain the rock edge point cloud. Then, an improved point cloud region growth method is proposed to solve the problem of large rock particles being overly divided into small particles. Finally, the rock particle size composition is analyzed based on the point cloud segmentation results. The proposed method is verified using rock piles mined by open-pit blasting. The results show that, compared with the sieving experiment, the proposed method has an average absolute error of less than 5%, saving time and manpower. In addition, the proposed method is based on the spatial distribution characteristics of point clouds and does not rely on deep learning, avoiding the complex data set production and model training process. The calculation process is interpretable and has broad application prospects. • Realized point cloud-based rock pile particle size analysis. • Proposed a rock particle point cloud edge detection algorithm. • Proposed a rock particle point cloud area growth algorithm. • Particle size analysis error is less than 5%.