BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Infimal convolution of Total Generalized Variation functionals for
  spatio-temporal regularization of image sequences - Martin Holler (Univer
 sity of Graz)
DTSTART:20150311T160000Z
DTEND:20150311T170000Z
UID:TALK57874@talks.cam.ac.uk
CONTACT:Davide Piazzoli
DESCRIPTION:Variational methods for image processing heavily rely on appro
 priate regularization functionals. While this topic is well investigated i
 n the still image context\, the question of suitable regularization for im
 age sequences is still quite open\, but not less important. In this talk\,
  we present a new approach for spatio-temporal regularization of image seq
 uences. When considering for instance the spatio-temporal Total Variation 
 (TV) or Total Generalized Variation (TGV) functional\, the scale of space 
 with respect to time is not given a-priori and in fact defines a trade-off
  between spatial and temporal regularization. This can be exploited to fur
 ther improve reconstruction quality by optimally balancing between two dif
 ferent scales via the infimal convolution of such functionals (ICTGV). We 
 present the analysis of the resulting regularization term and its applicat
 ion for dynamic MRI reconstruction and the artifact-free decompression of 
 MPEG compressed videos.
LOCATION:MR14\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR