BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Hyperanalytic Denoising - Sofia Olhede\, Dept. of Mathematics\, Im
 perial College London
DTSTART:20070412T120000Z
DTEND:20070412T130000Z
UID:TALK6713@talks.cam.ac.uk
CONTACT:Taylan Cemgil
DESCRIPTION:Image estimation requires a compromise between reconstruction 
 flexibility and computational tractability. A commonly adopted approach to
  the estimation problem is to decompose the observed image in a suitable b
 asis\, where the compression of the decomposition simplifies subsequent an
 alysis. Much effort has been expanded in designing appropriate 2-D decompo
 sitions\, but of great importance is also the statistical estimation proce
 dure chosen\, and this talk will focus on the estimation of any set of dec
 omposition coefficients.\n\nA new estimation method that can be combined w
 ith a local decomposition method is introduced. Under the assumption that 
 structured features correspond to highly concentrated and connected region
 s of the spatial and spatial frequency space additional image replicates w
 ith the same local structure as the observed image are constructed from th
 e observed image. The decomposition of the image is estimated not only usi
 ng the observed image decomposition coefficients\, but also using a set of
  local coefficients constructed from the replicate images. Given the tract
 able form of the first and second order structure of the full set of decom
 position coefficients of both the image and replicate images at any given 
 scale and spatial position\, the full procedure can be specified analytica
 lly\, and its risk calculated explicitly. The proposed method is implement
 ed on several examples\, and theoretical risk calculations substantiated\,
  as well as visually appealing reconstructions presented. \n\nThe slides o
 f the seminar are available at\nhttp://stats.ma.ic.ac.uk/sco/public_html/s
 eminarsIC.htm\n\nThis work was part of an EPSRC supported project.
LOCATION:LR5\, Engineering\, Department of
END:VEVENT
END:VCALENDAR