Boris Sliz
Supervisor(s): Martin Madaras
Slovak Technical University
Abstract: The development of 3D scanners in recent years enabled object scanning with high precision. In the process of 3D reconstruction, the first step is to scan the object with different points of view, while these scans are aligned. Despite the significant advances of 3D scanners, their acquisition is still costly, and capturing correct scans is time-consuming. In this paper, we focus on the simulation of the process of scan acquisition, alignment, and reconstruction in the virtual environment. Our work consists of the implementation of multiple approaches to simulate 3D scanning. We use a simulated robotic hand manipulator to achieve points where we capture virtual scans. Reconstructed virtual scans can provide information on whether we captured all parts of the scanning object, or we need additional data. We also simulate a realistic spawn of multiple parts into the bin, where we use the virtual scanner to capture the scene. Virtual scans of randomly spawned objects are useful as testing data for localization purposes.
Keywords: 3D Reconstruction, Computer Vision
Full text: Year: 2020
