Barsukov, Y. (2013). Development of GPICP, a GPU-only registration algorithm based on iterative closest point [Diploma Thesis]. reposiTUm. https://doi.org/10.34726/hss.2013.21784
E376 - Institut für Automatisierungs- und Regelungstechnik
-
Date (published):
2013
-
Number of Pages:
57
-
Keywords:
GPGPU; ICP; Iterative Closest Point; Registrierung; CUDA
de
GPGPU; ICP; Iterative Closest Point; registration; CUDA
en
Abstract:
The ICP algorithm, developed by Besl and McKay, has long been the -golden standard- for solving the registration problem. With the recent increase in programmability of GPUs, their parallel programming model has become a perfect medium for ICP. The goal of this work was to develop an ICP implementation that could run entirely on the GPU, solving the problems imposed by the GPU architecture and harvesting increased performance in an effort to make the registration more accurate.
en
Additional information:
Abweichender Titel laut Übersetzung der Verfasserin/des Verfassers Zsfassung in dt. Sprache
