Title:	COMBINATION OF SHAPE AND VISUAL INFORMATION FOR THE REGISTRATION OF 3D POINT CLOUDS FROM TOF CAMERAS
Author(s):	Dominik Aufderheide, Peter Planert, Werner Krybus, Gerard Edwards
ISBN:	978-972-8939-89-2
Editors:	Yingcai Xiao
Year:	2013
Edition:	Single
Keywords:	Point Cloud Registration (PCR), Time-of-Flight camera (ToF), Iterative Closest Point (ICP), RGB-D sensors
Type:	Full Paper
First Page:	27
Last Page:	34
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Computing the correct alignment of 3D point clouds is an import task for many different applications. The vast majority of the suggested procedures realise the spatial frame-to-frame alignment of the 3D measurements by applying the iterative closest point algorithm (ICP) or variants thereof. ICP in general considers only the 3D shape for the computation of the relative pose between two given point clouds. By using actual state-of-the-art time of flight sensors (ToF) there are also visual measurements available which are completely neglected within classical ICP. This paper describes a novel framework which also employs visual information as an aiding modality for the registration process, which results in a higher accuracy of the pose estimation and lower computational costs, in comparison to the classical ICP.
	Go Back