BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Below the Surface of the Non-Local Bayesian Image Denoising Method
  - Mila Nikolova (CNRS (Centre national de la recherche scientifique)\; EN
 S de Cachan)
DTSTART:20171101T090000Z
DTEND:20171101T095000Z
UID:TALK94249@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:joint work with Pablo Arias CMLA\, ENS Cachan\, CNRS\, Univers
 it&#19\;y Paris-Saclay  The non-local Bayesian (NLB) patch-based approach 
 of Lebrun\, Buades\, and Morel [1] is considered as a state-of-the-art met
 hod for the restoration of (color) images corrupted by white Gaussian nois
 e. It gave rise to numerous ramiifications like e.g.\, possible improvemen
 ts\, processing of various data sets and video. This work is the first att
 empt to analyse the method in depth in order to understand the main phenom
 ena underlying its effectiveness. Our analysis\, corroborated by numerical
  tests\, shows several unexpected facts. In a variational setting\, the fi
 rst-step Bayesian approach to learn the prior for patches is equivalent to
  a pseudo-Tikhonov regularisation where the regularisation parameters can 
 be positive or negative. Practically very good results in this step are ma
 inly due to the aggregation stage - whose importance needs to be re-evalua
 ted.  Reference [1] Lebrun\, M.\, Buades\, A.\, Morel\, J.M.: A nonlocal B
 ayesian image denoising algorithm. SIAM J. Imaging Sci.6(3)\, 1665-1688 (2
 013)
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR