Statistics – Applications
Scientific paper
Jan 2009
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2009aas...21342704p&link_type=abstract
American Astronomical Society, AAS Meeting #213, #427.04; Bulletin of the American Astronomical Society, Vol. 41, p.258
Statistics
Applications
Scientific paper
Effective analysis of large multispectral and multitemporal data sets demands new ways of data representation. We present applications of standard and original methods of data dimension reduction to astrophysical images (finding significant low-dimensional structures concealed in high-dimensional data). Such methods are widely used already outside of astronomy to effectively analyze large data sets. Data dimension reduction facilitates data organization, retrieval, and analysis (by improving statistical inference), which are crucial to multiwavelength astronomy, archival research, large-scale digital sky surveys and temporal astronomy. These methods allow a user to reduce a large number of FITS images, e.g. each representing a different wavelength, into a few images retaining more than 95% of the original visual information. An immediate simple application of this would be creating a multiwavelength "quick-look" image that includes all essential information in a statistically justified way, and thus is much more accurate than a "quick-look" made by simple coadding with an ad hoc, heuristic weighting. The dimensionally-reduced image is also naturally much smaller in file size in bytes than the total summed size of the non-dimensionally-reduced images. Thus dimensionally-reduced images offer an enormous savings in storage space and database-transmission bandwidth for the user. An analogous process of dimension reduction is possible for a large set of images obtained at the same wavelength but at different times (e.g. LSST images). Other applications of data dimension reduction include, but are not limited to, decorrelating data elements, removing noise, artifact separation, feature extraction, clustering and pattern classification in astronomical images. We demonstrate applications of the algorithms to test cases of current space-based IR data from the Spitzer Space Telescope.
Ardila David
Carey Sean
Ingalls James
McCollum Bruce
Pesenson Isaac
No associations
LandOfFree
More to Astronomical Images than Meets the Eye: Data Dimension Reduction for Efficient Data Organization, Retrieval and Advanced Visualization and Analysis of Large Multitemporal/Multispectral Data Sets 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 More to Astronomical Images than Meets the Eye: Data Dimension Reduction for Efficient Data Organization, Retrieval and Advanced Visualization and Analysis of Large Multitemporal/Multispectral Data Sets, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and More to Astronomical Images than Meets the Eye: Data Dimension Reduction for Efficient Data Organization, Retrieval and Advanced Visualization and Analysis of Large Multitemporal/Multispectral Data Sets will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1703134