||Evaluation of surface registration algorithms for PET motion correction
||Kjer, Hans Martin
||Paulsen, Rasmus Reinhold (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||Recent advancements in high resolution Positron Emission Tomography (PET)
have made it possible to study metabolic activity and the anatomy of the brain
in great details. It becomes increasingly important to perform motion correction
on these data. Motion tracking with a structured light system is pursued by
Ph.D. student Oline V. Olesen.
This thesis covers the registration (alignment) of point cloud surfaces acquired
from a structured light scanner.
An introduction to the concepts of rigid motion and point cloud properties is
given. These include representations of rotations, surface curvature estimates
and solutions to the nearest neighbor problem. The iterative closest point (ICP)
algorithm is detailed with some of its most promising variants. Non linear
optimization is introduced conceptually.
Experimental data is used to compare these different variants in terms of convergence,
accuracy, robustness and speed.
It is concluded, that implementational simple variations of the ICP algorithm
can yield much improved performance and reliability, and that, with the proper
choice of acceleration techniques, real time performance is a realistic perspective.
In the appendix derivations of the most important mathematical formulas are
||Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Creation date: 2010-06-02
Update date: 2010-06-02