File(s) under embargo
until file(s) become available
Developing a system for rendering a small object: capture, segmentation, camera calibration, and 3D reconstruction
thesisposted on 03.12.2020, 16:18 by Yin Wang
As inkjet printing technology has matured, there are opportunities for novel applications of inkjet printers. Specifically, the printer could be a convenient tool for people who want to print some colorful patterns on a non-flat smooth surface.
In this dissertation, a low-cost, single camera, double-mirror system that can be built in a desktop printer will be described. The usage of this system is to capture the image of a small object and then to print a graphic on it. The object has a smooth non-flat surface. The rendering process will include image segmentation, camera calibration, and visual hull 3D reconstruction. In each image processing phase, a novel algorithm will be developed to refine the result. We will state the mathematical principles behind each phase, and also show the mechanical set up for the whole system. Experimental results will be shown as well, which illustrate the accuracy of our system to handle the 3D reconstruction task.