| dc.description.abstract | Emerging advances in robotic cameras, long-range wireless networking, and
distributed sensors make feasible a new class of hybrid teleoperated/autonomous robotic
remote "observatories" that can allow groups of peoples, via the Internet, to observe,
record, and index detailed activity occurred in remote site. Equipped with robotic
pan-tilt actuation mechanisms and a high-zoom lens, the camera can cover a large
region with very high spatial resolution and allows for observation at a distance.
High resolution motion panorama is the most nature data representation. We develop algorithms and protocols for high resolution motion panorama. We discover and prove the projection invariance and achieve real time image alignment. We propose a minimum variance based incremental frame alignment algorithm to minimize the accumulation of alignment error in incremental image alignment and ensure
the quality of the panorama video over the long run. We propose a Frame Graph based panorama documentation algorithm to manage the large scale data involved in the online panorama video documentation. We propose a on-demand high resolution panorama video-streaming system that allows on-demand sharing of a high-resolution motion panorama and efficiently deals with multiple concurrent spatial-temporal user requests. In conclusion, our research work on high resolution motion panorama have significantly improve the efficiency and accuracy of image alignment, panorama video
quality, data organization, and data storage and retrieving in remote observation using networked robotic cameras. | en |
