Title :
An iterative framework for projection-based image sequence registration
Author_Institution :
CICATA-IPN, Queretaro, Mexico
Abstract :
In this paper, an iterative framework for projection-based image sequence registration to be used by a non-holonomic mobile robot is introduced. Since obtaining complete registration has shown to be difficult and error prone, it is claimed that it is worth pursuing trying to gather at least partial and qualitative information from projections. A tracking method is adapted to select and track features between projections. Some results are shown with sequences of images taken when a mobile robot was heading forward, approximately along its optical axis, and rotating approximately around its optical center. It is shown that is possible to interpret camera motion from the projection of individual frames in an image stream.
Keywords :
image registration; image sequences; iterative methods; mobile robots; motion estimation; optical tracking; robot vision; camera motion; error prone; image stream; iterative framework; least partial information; nonholonomic mobile robot; optical axis; optical center; projection based image sequence registration; qualitative information; tracking method; Biomedical optical imaging; Cameras; Computer vision; Geometrical optics; Image sequences; Mobile robots; Optical sensors; Robot vision systems; Robustness; Streaming media;
Conference_Titel :
Robotics and Automation, 2003. Proceedings. ICRA '03. IEEE International Conference on
Print_ISBN :
0-7803-7736-2
DOI :
10.1109/ROBOT.2003.1241871