Registration of point cloud data for matching crushed sand particles

Abstract

Identifying the fracture surface of fragments that are resulted from the mechanical crushing of a single sand particle and matching individual fragments to the mother particle remains a challenge in the geomechanics research. This paper presents a novel particle tracking method that can track fractured fragments from the original sand particle through the application of the point cloud registration technique based on the X-ray tomography data. Firstly, the external morphologies of the mother particle and the child particles were acquired from the X-ray tomography data from the single particle crushing test. A series of imaging process techniques were applied to treat the raw images and obtain the particle fracture pattern. Secondly, after the image processing, the contour-based mesh segmentation algorithm was utilised to divide a child particle into multiple faces. Then the iterative closest point algorithm was used to try matching each face of the child particle to the mother particle and evaluate the matching results through the distance error. Finally, upon a successful match, the region with a higher point density on the child particle can be extracted and marked as a part of the original particle surface, and the rest part marked as a part of the fracture surface. The effectiveness and efficiency of the tracking method were demonstrated using the tomography data of 9 crushed Leighton Buzzard sand particles.