Please use this identifier to cite or link to this item:
Synchronization of an inertial measurement unit and standalone camera system based on attitude and optical flow data
|Arnold_Lee_r.pdf||Version for non-UH users. Copying/Printing is not permitted||10.28 MB||Adobe PDF||View/Open|
|Arnold_Lee_uh.pdf||Version for UH users||10.27 MB||Adobe PDF||View/Open|
|Title:||Synchronization of an inertial measurement unit and standalone camera system based on attitude and optical flow data|
|Authors:||Arnold, Lee Anthony|
inertial measurement unit
standalone camera system
|Issue Date:||Aug 2012|
|Publisher:||[Honolulu] : [University of Hawaii at Manoa], [August 2012]|
|Abstract:||The fusion of camera and inertial data is well established and useful for a wide range of tasks. With advancements in camera technology and renewed interest in computer vision, standalone cameras have emerged as a possible research tool. These cameras offer high quality video recording, in a small, lightweight form factor, but are limited by connectivity, thus limiting their usefulness. This thesis presents a software correlation method which correlates inertial data and camera data based on attitude estimates, eliminating the need for additional hardware or connections.|
This thesis presents a brief background and description of the correlation process, along with a review of similar work. The underlying methods for feature detection, optical flow, and signal comparison are explained in detail in the context of the software correlation method. The software correlation method is then demonstrated in an experiment that looks at a single axis of rotation. Finally, the error in the software correlation process is quantified and compared to that in an uncorrelated system. Software correlation proves to be a viable option for the correlation of inertial measurement units and standalone cameras.
|Description:||M.S. University of Hawaii at Manoa 2012.|
Includes bibliographical references.
|Appears in Collections:||M.S. - Mechanical Engineering|
Items in ScholarSpace are protected by copyright, with all rights reserved, unless otherwise indicated.