Edge detection in unorganized 3D point cloud

Abstract

The application of 3D laser scanning in the mining industry is increasing progressively over the years. This presents an opportunity to visualize and analyze the underground world and potentially save countless man- hours and exposure to safety incidents. This thesis envisions to detect the “Edges of the Rocks” in the 3D point cloud collected via scanner, although edge detection in point cloud is considered as a difficult but meaningful problem. As a solution to noisy and unorganized 3D point cloud, a new method, EdgeScan method, has been proposed and implemented to detect fast and accurate edges from the 3D point cloud for real time systems. EdgeScan method is aimed to make use of 2D edge processing techniques to represent the edge characteristics in 3D point cloud with better accuracy. A comparisons of EdgeScan method with other common edge detection methods for 3D point cloud is administered, eventually, results suggest that the stated EdgeScan method furnishes a better speed and accuracy especially for large dataset in real time systems.