Title:	A FRAMEWORK FOR REAL TIME SCENE MODELING BASED ON VISUAL-INERTIAL CUES
Author(s):	Dominik Aufderheide, Werner Krybus
ISBN:	978-972-8939-22-9
Editors:	Yingcai Xiao, Tomaz Amon and Piet Kommers
Year:	2010
Edition:	Single
Keywords:	Structure from Motion (SfM), Sensor Fusion, Inertial Sensing, Strapdown algorithm, Scene Modeling
Type:	Short Paper
First Page:	385
Last Page:	390
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	The self-acting generation of three-dimensional scene or object models is a highly desirable goal for numerous applications in different branches as paramedics, manufacturing, engineering or entertainment. Nowadays almost all commercial systems are based on specialized and partially fixed hardware, which is only available in upper price segments. This paper describes a framework for the development of a hand-held scene acquisition device which is based on standardized low-cost inertial and vision sensors (Visual-inertial Scene Acquisition - VISA). The combination of inertial and optical modalities is used due to the complementary characteristics of both sensor types. Such a sensor fusion concept is able to outperform classical Structure from Motion (SfM) algorithms in the field of computer vision in terms of robustness and computational costs.
	Go Back