BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Sense and nonsense of box-constrained total variation minimization
  - Andreas Langer\, Universität Stuttgart
DTSTART:20160729T150000Z
DTEND:20160729T160000Z
UID:TALK66927@talks.cam.ac.uk
CONTACT:Martin Benning
DESCRIPTION:A good approximation of the original image from an observed im
 age may be obtained by minimising a functional that consists of a data-fid
 elity term\, a regularisation term\, and a parameter\, which balances data
 -fidelity and regularisation. Using the total variation as a regularisatio
 n term is a rather well understood concept of restoring images while prese
 rving edges and discontinuities. If we have knowledge about the dynamic ra
 nge in which the original image lies\, then it seems natural to incorporat
 e this information (via a box-constraint) into the model. Moreover\, it is
  clear that the minimiser of the considered functional highly depends on t
 he proper choice of the parameter.\nIn this talk we are wondering whether 
 incorporating a box-constraint into the model really improves the quality 
 of the solution or if it is indeed more a question of the proper choice of
  the regularisation parameter. Further we propose a semi-smooth Newton met
 hod for solving the considered models.
LOCATION:MR 14\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR