Computer Science – Learning
Scientific paper
Dec 2002
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2002agufmsh51a0433s&link_type=abstract
American Geophysical Union, Fall Meeting 2002, abstract #SH51A-0433
Computer Science
Learning
2407 Auroral Ionosphere (2704), 2494 Instruments And Techniques, 9350 North America, 9820 Techniques Applicable In Three Or More Fields
Scientific paper
Modern ground-based auroral imaging instruments produce vast amounts of data. Increased temporal and spatial resolution allow for more detailed studies, but typically only a fraction of the data is utilised due to the extreme manual labour required for analysing all images. Full analysis of all images is only possible through the use of machine vision, in which classification of image contents is performed automatically. Our recent research has resulted in a content-based image retrieval system, which can be used to locate similar auroral images in the CANOPUS all-sky image set consisting of over 200000 images. The retrieval is initiated by a user supplied image, after which supervised learning techniques are utilised to refine the search for similar images. Based on the automatically classified auroral data, we discuss the occurrence of arcs and patchy aurora within the context of auroral precipitation boundaries, and the local time distribution of other space physical phenomena which are likely related to the formation of these auroral shapes (eg., Pc5 pulsations).
Donovan Eric F.
Syrjaesuo M. T.
No associations
LandOfFree
Auroral Image Classification via Machine Vision does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Auroral Image Classification via Machine Vision, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Auroral Image Classification via Machine Vision will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1436317