Point cloud merging for complete 3D surface reconstruction

Abstract

This article addresses the problem of reconstructing 3D surfaces from unorganized point sets (also known as point clouds). This issue is common to many different areas of science and engineering, including computer graphics and computer vision. Unorganized point sets can be acquired in different ways, e.g., using laser scanners, computer tomography, magnetic resonance imaging, multicamera vision systems, etc. This paper presents a method for point cloud merging that allows calculating the initial rotation and translation between point clouds of different viewing angles. By tracking the marker's position, scanning and point cloud registration can be done in real time and with high accuracy.